Apple M3 Ultra (512 GB) vs Apple M1 (16 GB) for local AI
Pick two to four systems or devices to see which runs more models, faster, for the kind of work you do.
Winner: Apple M3 Ultra (512 GB)
For general chat: Its ten best models average a score of 100 vs 82; it runs 236 more LLMs (616 vs 380); it is about 1.6× faster on its top models.
 Apple M3 Ultra (512 GB)
|
 Apple M1 (16 GB)
|
|
|---|---|---|
| Memory | 512 GB unified | 16.0 GB unified |
| Usable for models | 384 GB | 10.7 GB |
| Memory bandwidth | 819 GB/s | 68 GB/s |
| FP16 compute | 28.4 TFLOPS | 2.6 TFLOPS |
| Launch price | — | — |
| Released | 2025 | 2020 |
| LLMs that run | 616 of 622 | 380 of 622 |
| LLMs that run well | 600 | 307 |
| Image models that run | 61 of 61 | 53 of 61 |
| Video models that run | 43 of 43 | 12 of 43 |
| Decision models that run | 70 of 70 | 64 of 70 |
| Top-10 LLM average score | 100 | 82 |
| Top-10 LLM average speed | 40.1 tok/s | 24.6 tok/s |
| Best LLM | MiniMax M3.1 Flash (428B-A23B) | Ling 3.0 Tiny (8B-A1.3B) |
Highlighted values are the best of the two. Model results assume 8K of context and the default macOS GPU memory limit on Macs.
LLMs
616 of 622 LLMs run on the Apple M3 Ultra (512 GB). 384 GB usable for models.
380 of 622 LLMs run on the Apple M1 (16 GB). 10.7 GB usable for models.
| Model | QualityM3 Ultra | SpeedM3 Ultra | RunsM3 Ultra | QualityM1 | SpeedM1 | RunsM1 |
|---|---|---|---|---|---|---|
| MiniMax M3.1 Flash (428B-A23B) | 92Excellent quality (A+) · 92 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 100/100 (quality, fit, speed and memory use) sets the ranking. | 33.4 tok/s | Runs great | 67Good quality (B) · 67 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Nex N2.6 (321B-A18B) | 92Excellent quality (A+) · 92 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 100/100 (quality, fit, speed and memory use) sets the ranking. | 41.4 tok/s | Runs great | 67Good quality (B) · 67 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| GLM-5.3 Flash (320B-A18B) | 92Excellent quality (A+) · 92 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 100/100 (quality, fit, speed and memory use) sets the ranking. | 41.4 tok/s | Runs great | 67Good quality (B) · 67 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Intern-S2 (397B-A17B) | 92Excellent quality (A+) · 92 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 100/100 (quality, fit, speed and memory use) sets the ranking. | 36.7 tok/s | Runs great | 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Nex N2 Pro (397B-A17B) | 92Excellent quality (A+) · 92 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 100/100 (quality, fit, speed and memory use) sets the ranking. | 36.7 tok/s | Runs great | 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen3.5 397B-A17B | 92Excellent quality (A+) · 92 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 100/100 (quality, fit, speed and memory use) sets the ranking. | 36.7 tok/s | Runs great | 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Ornith 1.5 397B-A17B | 91Excellent quality (A+) · 91 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 100/100 (quality, fit, speed and memory use) sets the ranking. | 36.7 tok/s | Runs great | 65Good quality (B) · 65 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| MiniMax M2.7 (230B-A10B) | 92Excellent quality (A+) · 92 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 99/100 (quality, fit, speed and memory use) sets the ranking. | 49.4 tok/s | Runs great | 65Good quality (B) · 65 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| MiMo V2 Flash (309B-A15B) | 88Excellent quality (A+) · 88 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 99/100 (quality, fit, speed and memory use) sets the ranking. | 40.9 tok/s | Runs great | 63Good quality (B) · 63 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| MiMo V2.6 Flash (310B-A15B) | 91Excellent quality (A+) · 91 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 99/100 (quality, fit, speed and memory use) sets the ranking. | 48.3 tok/s | Runs great | 67Good quality (B) · 67 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| MiniMax M2.5 (230B-A10B) | 91Excellent quality (A+) · 91 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 99/100 (quality, fit, speed and memory use) sets the ranking. | 49.4 tok/s | Runs great | 64Good quality (B) · 64 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Inkling Small (276B-A12B) | 90Excellent quality (A+) · 90 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 99/100 (quality, fit, speed and memory use) sets the ranking. | 49.3 tok/s | Runs great | 65Good quality (B) · 65 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| K2-Horizon (375B-A23B) | 89Excellent quality (A+) · 89 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 99/100 (quality, fit, speed and memory use) sets the ranking. | 33.4 tok/s | Runs great | 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Naive N0.5 Flash (309B-A15.5B) | 90Excellent quality (A+) · 90 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 98/100 (quality, fit, speed and memory use) sets the ranking. | 47.0 tok/s | Runs great | 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| MiniMax M2.1 (230B-A10B) | 90Excellent quality (A+) · 90 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 98/100 (quality, fit, speed and memory use) sets the ranking. | 49.4 tok/s | Runs great | 63Good quality (B) · 63 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Ling 3.1 Flash 1019 (560B-A25B) | 92Excellent quality (A+) · 92 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 98/100 (quality, fit, speed and memory use) sets the ranking. | 31.0 tok/s | Runs welltight fit | 68Good quality (B) · 68 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen3.8 Flash Next (125B-A6B) | 95Excellent quality (A+) · 95 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 98/100 (quality, fit, speed and memory use) sets the ranking. | 73.5 tok/s | Runs great | 67Good quality (B) · 67 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| MiniMax M3 (427B) | 90Excellent quality (A+) · 90 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 98/100 (quality, fit, speed and memory use) sets the ranking. | 31.0 tok/s | Runs great | 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Step 3.7 Flash (198B-A11B) | 92Excellent quality (A+) · 92 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 98/100 (quality, fit, speed and memory use) sets the ranking. | 45.6 tok/s | Runs great | 65Good quality (B) · 65 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| MiMo V2.5 (310B-A15B) | 89Excellent quality (A+) · 89 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 98/100 (quality, fit, speed and memory use) sets the ranking. | 48.3 tok/s | Runs great | 65Good quality (B) · 65 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| DeepSeek V4 Flash (284B-A13B) | 89Excellent quality (A+) · 89 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 98/100 (quality, fit, speed and memory use) sets the ranking. | 54.4 tok/s | Runs great | 65Good quality (B) · 65 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Ling 3.1 Flash (560B-A25B) | 90Excellent quality (A+) · 90 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 98/100 (quality, fit, speed and memory use) sets the ranking. | 31.0 tok/s | Runs welltight fit | 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| MiniMax M2 (230B-A10B) | 88Excellent quality (A+) · 88 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 97/100 (quality, fit, speed and memory use) sets the ranking. | 49.4 tok/s | Runs great | 62Good quality (B) · 62 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Beam 1.0 (501B-A23B) | 89Excellent quality (A+) · 89 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 97/100 (quality, fit, speed and memory use) sets the ranking. | 32.9 tok/s | Runs welltight fit | 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Step 3.5 Flash (196B-A11B) | 91Excellent quality (A+) · 91 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 97/100 (quality, fit, speed and memory use) sets the ranking. | 45.6 tok/s | Runs great | 64Good quality (B) · 64 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Hunyuan Hy3 (295B-A21B) | 89Excellent quality (A+) · 89 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 97/100 (quality, fit, speed and memory use) sets the ranking. | 36.2 tok/s | Runs great | 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Kolibri 1 (78B-A3.5B) | 91Excellent quality (A+) · 91 of 100: how capable the model is at F16, whatever it runs on. Match score 97/100 (quality, fit, speed and memory use) sets the ranking. | 46.7 tok/s | Runs great | 63Good quality (B) · 63 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Solar Open 2 250B | 87Excellent quality (A+) · 87 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 96/100 (quality, fit, speed and memory use) sets the ranking. | 40.9 tok/s | Runs great | 63Good quality (B) · 63 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| GLM-4.7 (355B-A32B) | 88Excellent quality (A+) · 88 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 96/100 (quality, fit, speed and memory use) sets the ranking. | 24.8 tok/s | Runs well | 65Good quality (B) · 65 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| DeepSeek V4 Flash (April 2026) | 91Excellent quality (A+) · 91 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 95/100 (quality, fit, speed and memory use) sets the ranking. | 39.6 tok/s | Runs great | 64Good quality (B) · 64 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen3-Next 80B-A3B Thinking | 86Excellent quality (A+) · 86 of 100: how capable the model is at F16, whatever it runs on. Match score 95/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 60Good quality (B) · 60 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Step 5 Preview (600B-A27B) | 81Great quality (A) · 81 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 95/100 (quality, fit, speed and memory use) sets the ranking. | 36.5 tok/s | Runs welltight fit | 67Good quality (B) · 67 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Nemotron 3.5 Super VL 120B-A12B | 92Excellent quality (A+) · 92 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 94/100 (quality, fit, speed and memory use) sets the ranking. | 42.4 tok/s | Runs great | 65Good quality (B) · 65 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen3 Next 80B-A3B Instruct | 84Great quality (A) · 84 of 100: how capable the model is at F16, whatever it runs on. Match score 94/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 58Good quality (B) · 58 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Ornith 1.5 122B-A10B | 91Excellent quality (A+) · 91 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 94/100 (quality, fit, speed and memory use) sets the ranking. | 49.4 tok/s | Runs great | 64Good quality (B) · 64 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| LongCat Flash Chat (560B) | 83Great quality (A) · 83 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 94/100 (quality, fit, speed and memory use) sets the ranking. | 28.9 tok/s | Runs welltight fit | 61Good quality (B) · 61 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen3.5 122B-A10B | 91Excellent quality (A+) · 91 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 94/100 (quality, fit, speed and memory use) sets the ranking. | 49.4 tok/s | Runs great | 64Good quality (B) · 64 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| NorthCUA Pro | 75Great quality (A) · 75 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 93/100 (quality, fit, speed and memory use) sets the ranking. | 36.7 tok/s | Runs great | 54Fair quality (C) · 54 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen3.6 35B-A3B | 93Excellent quality (A+) · 93 of 100: how capable the model is at F16, whatever it runs on. Match score 93/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 65Good quality (B) · 65 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Nemotron 3.5 Bolt 74B-A9B | 92Excellent quality (A+) · 92 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 93/100 (quality, fit, speed and memory use) sets the ranking. | 41.3 tok/s | Runs great | 64Good quality (B) · 64 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| MiniCPM-V 4.7 35B-A3B | 93Excellent quality (A+) · 93 of 100: how capable the model is at F16, whatever it runs on. Match score 93/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 65Good quality (B) · 65 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen3.5 35B-A3B | 93Excellent quality (A+) · 93 of 100: how capable the model is at F16, whatever it runs on. Match score 93/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 65Good quality (B) · 65 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen3 235B-A22B Instruct 2507 | 85Excellent quality (A+) · 85 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 93/100 (quality, fit, speed and memory use) sets the ranking. | 34.7 tok/s | Runs great | 63Good quality (B) · 63 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Nemotron 3 Ultra 550B-A55B | 88Excellent quality (A+) · 88 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 93/100 (quality, fit, speed and memory use) sets the ranking. | 14.9 tok/s | Runs welltight fit | 65Good quality (B) · 65 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| GLM-4.6 (355B-A32B) | 79Great quality (A) · 79 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 92/100 (quality, fit, speed and memory use) sets the ranking. | 24.8 tok/s | Runs well | 58Good quality (B) · 58 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen3-VL 235B-A22B Instruct | 84Great quality (A) · 84 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 92/100 (quality, fit, speed and memory use) sets the ranking. | 34.7 tok/s | Runs great | 62Good quality (B) · 62 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Nex N2 Mini (35B-A3B) | 92Excellent quality (A+) · 92 of 100: how capable the model is at F16, whatever it runs on. Match score 92/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 64Good quality (B) · 64 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen3 235B-A22B | 83Great quality (A) · 83 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 92/100 (quality, fit, speed and memory use) sets the ranking. | 34.7 tok/s | Runs great | 61Good quality (B) · 61 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Nemotron 3 Super 120B-A12B | 87Excellent quality (A+) · 87 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 92/100 (quality, fit, speed and memory use) sets the ranking. | 42.4 tok/s | Runs great | 61Good quality (B) · 61 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Flint (40B-A2.5B) | 90Excellent quality (A+) · 90 of 100: how capable the model is at F16, whatever it runs on. Match score 92/100 (quality, fit, speed and memory use) sets the ranking. | 59.9 tok/s | Runs great | 63Good quality (B) · 63 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Sage-A1.5 35B-A3B | 91Excellent quality (A+) · 91 of 100: how capable the model is at F16, whatever it runs on. Match score 92/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 63Good quality (B) · 63 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| K-EXAONE 236B-A23B | 83Great quality (A) · 83 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. | 33.4 tok/s | Runs great | 61Good quality (B) · 61 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Ling 3.0 Flash (127B-A5B) | 82Great quality (A) · 82 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. | 65.0 tok/s | Runs great | 57Good quality (B) · 57 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| DeepSeek V3.2 (671B) | 78Great quality (A) · 78 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. | 27.5 tok/s | Runs welltight fit | 65Good quality (B) · 65 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| GLM-4.5 (355B-A32B) | 77Great quality (A) · 77 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. | 24.8 tok/s | Runs well | 57Good quality (B) · 57 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Ornith 1.5 35B-A3B | 90Excellent quality (A+) · 90 of 100: how capable the model is at F16, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 63Good quality (B) · 63 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Gemma 4 26B-A4B | 92Excellent quality (A+) · 92 of 100: how capable the model is at F16, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. | 43.3 tok/s | Runs great | 64Good quality (B) · 64 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 24.1 tok/s | Slowfar over GPU memory limit |
| DeepSeek V3.2 Exp | 77Great quality (A) · 77 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. | 27.5 tok/s | Runs welltight fit | 64Good quality (B) · 64 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Apodex 1.1 Mini (35B-A3B) | 89Excellent quality (A+) · 89 of 100: how capable the model is at F16, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 62Good quality (B) · 62 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Ornith 1.0 35B-A3B | 89Excellent quality (A+) · 89 of 100: how capable the model is at F16, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 62Good quality (B) · 62 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| DeepSeek R1 (671B) | 77Great quality (A) · 77 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. | 27.5 tok/s | Runs welltight fit | 64Good quality (B) · 64 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| DeepSeek V3.1 Terminus | 76Great quality (A) · 76 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. | 27.5 tok/s | Runs welltight fit | 64Good quality (B) · 64 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| DeepSeek V3.1 (671B) | 76Great quality (A) · 76 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. | 27.5 tok/s | Runs welltight fit | 63Good quality (B) · 63 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Llama 4 Maverick (17B x 128E) | 69Good quality (B) · 69 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. | 36.7 tok/s | Runs great | 50Fair quality (C) · 50 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Mistral Small 4 (119B-A6B) | 80Great quality (A) · 80 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. | 57.4 tok/s | Runs great | 56Good quality (B) · 56 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Command A+ (218B-A25B) | 83Great quality (A) · 83 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. | 31.0 tok/s | Runs great | 61Good quality (B) · 61 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Baichuan M3 235B-A22B | 79Great quality (A) · 79 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. | 34.7 tok/s | Runs great | 58Good quality (B) · 58 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| MiniMax-M1 80K | 78Great quality (A) · 78 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. | 17.7 tok/s | Runs well | 58Good quality (B) · 58 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Agents-A1 35B-A3B | 87Excellent quality (A+) · 87 of 100: how capable the model is at F16, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 60Good quality (B) · 60 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| GLM-4.7 Flash (30B-A3B) | 88Excellent quality (A+) · 88 of 100: how capable the model is at F16, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 61Good quality (B) · 61 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen3 Coder 480B-A35B | 86Excellent quality (A+) · 86 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. | 22.8 tok/s | Runs well | 63Good quality (B) · 63 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen AgentWorld 35B-A3B | 87Excellent quality (A+) · 87 of 100: how capable the model is at F16, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 60Good quality (B) · 60 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen3 Coder Next (80B-A3B) | 86Excellent quality (A+) · 86 of 100: how capable the model is at F16, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 60Good quality (B) · 60 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| InternVL3.5 241B-A28B | 81Great quality (A) · 81 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. | 28.0 tok/s | Runs well | 60Good quality (B) · 60 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Trinity Large Thinking (400B-A13B) | 77Great quality (A) · 77 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. | 46.1 tok/s | Runs great | 56Good quality (B) · 56 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| MiniMax-M1 40K | 77Great quality (A) · 77 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. | 17.7 tok/s | Runs well | 57Good quality (B) · 57 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| DeepSeek V4.1 Flash 1017 (552B-A16B) | 68Good quality (B) · 68 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. | 69.5 tok/s | Runs welltight fit | 68Good quality (B) · 68 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Mistral Large 3 (675B) | 75Great quality (A) · 75 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. | 25.1 tok/s | Runs welltight fit | 63Good quality (B) · 63 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Ling 2.6 Flash (104B-A7.4B) | 80Great quality (A) · 80 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. | 48.6 tok/s | Runs great | 56Good quality (B) · 56 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| DeepSeek V4.1 Flash (552B-A16B) | 67Good quality (B) · 67 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. | 69.5 tok/s | Runs welltight fit | 67Good quality (B) · 67 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| GLM-4.6V | 82Great quality (A) · 82 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. | 42.4 tok/s | Runs great | 58Good quality (B) · 58 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| gpt-oss 120B | 84Great quality (A) · 84 of 100: how capable the model is at MXFP4, whatever it runs on. Match score 88/100 (quality, fit, speed and memory use) sets the ranking. | 113 tok/s | Runs great | 84Great quality (A) · 84 of 100: how capable the model is at MXFP4, whatever it runs on. Match score 11/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen3 235B-A22B Thinking 2507 | 86Excellent quality (A+) · 86 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 88/100 (quality, fit, speed and memory use) sets the ranking. | 34.7 tok/s | Runs great | 63Good quality (B) · 63 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| GLM-5.3 (753B-A40B) | 68Good quality (B) · 68 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 88/100 (quality, fit, speed and memory use) sets the ranking. | 32.0 tok/s | Runs welltight fit | 68Good quality (B) · 68 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Raon-1 mini Preview (30B-A4B) | 88Excellent quality (A+) · 88 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 88/100 (quality, fit, speed and memory use) sets the ranking. | 73.7 tok/s | Runs great | 61Good quality (B) · 61 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 21.6 tok/s | Slowfar over GPU memory limit |
| Laguna S 2.1 (118B-A8B) | 86Excellent quality (A+) · 86 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 88/100 (quality, fit, speed and memory use) sets the ranking. | 45.5 tok/s | Runs great | 60Good quality (B) · 60 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| GLM-4.5V | 79Great quality (A) · 79 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. | 42.4 tok/s | Runs great | 56Good quality (B) · 56 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Laguna XS 2.1 (33B-A3B) | 82Great quality (A) · 82 of 100: how capable the model is at F16, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 57Good quality (B) · 57 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| DeepSeek V3 0324 (671B) | 69Good quality (B) · 69 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. | 27.5 tok/s | Runs welltight fit | 58Good quality (B) · 58 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Leanstral 119B-A6B | 74Great quality (A) · 74 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. | 53.9 tok/s | Runs great | 51Fair quality (C) · 51 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| GLM-5.2 (753B-A40B) | 67Good quality (B) · 67 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. | 32.0 tok/s | Runs welltight fit | 67Good quality (B) · 67 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| dots.llm1 Instruct | 75Great quality (A) · 75 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. | 37.2 tok/s | Runs great | 53Fair quality (C) · 53 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Solar Open 100B | 79Great quality (A) · 79 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. | 42.4 tok/s | Runs great | 56Good quality (B) · 56 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Ling Flash 2.0 (100B-A6B) | 76Great quality (A) · 76 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. | 56.7 tok/s | Runs great | 53Fair quality (C) · 53 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| MiniMax-Text-01 | 72Great quality (A) · 72 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. | 17.7 tok/s | Runs well | 53Fair quality (C) · 53 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Muse Spark Open (816B-A42B) | 67Good quality (B) · 67 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. | 30.6 tok/s | Runs welltight fit | 67Good quality (B) · 67 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| GLM-5.1 (744B-A40B) | 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. | 32.0 tok/s | Runs welltight fit | 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| GLM-5 (744B-A40B) | 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. | 32.0 tok/s | Runs welltight fit | 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| GLM-4.5 Air (106B-A12B) | 77Great quality (A) · 77 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. | 42.4 tok/s | Runs great | 54Fair quality (C) · 54 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| ERNIE 4.5 300B-A47B | 76Great quality (A) · 76 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. | 17.3 tok/s | Runs well | 56Good quality (B) · 56 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| KAT-Coder V2.5 Dev (35B-A3B) | 91Excellent quality (A+) · 91 of 100: how capable the model is at F16, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 63Good quality (B) · 63 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen3.8 27B | 90Excellent quality (A+) · 90 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 86/100 (quality, fit, speed and memory use) sets the ranking. | 31.2 tok/s | Runs great | 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 3.6 tok/s | Slowfar over GPU memory limit |
| Nemotron Cascade 2 30B-A3B | 80Great quality (A) · 80 of 100: how capable the model is at F16, whatever it runs on. Match score 86/100 (quality, fit, speed and memory use) sets the ranking. | 46.3 tok/s | Runs great | 56Good quality (B) · 56 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| DeepSeek R1 0528 (671B) | 78Great quality (A) · 78 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 86/100 (quality, fit, speed and memory use) sets the ranking. | 27.5 tok/s | Runs welltight fit | 65Good quality (B) · 65 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen3.6 27B | 89Excellent quality (A+) · 89 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 86/100 (quality, fit, speed and memory use) sets the ranking. | 31.2 tok/s | Runs great | 65Good quality (B) · 65 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 3.6 tok/s | Slowfar over GPU memory limit |
| Nanbeige4.3 33B | 90Excellent quality (A+) · 90 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 86/100 (quality, fit, speed and memory use) sets the ranking. | 26.1 tok/s | Runs well | 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Kimi Linear 48B-A3B | 74Great quality (A) · 74 of 100: how capable the model is at F16, whatever it runs on. Match score 86/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 51Fair quality (C) · 51 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Hunyuan Hy4 Preview (780B) | 67Good quality (B) · 67 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 85/100 (quality, fit, speed and memory use) sets the ranking. | 23.9 tok/s | Runs welltight fit | 67Good quality (B) · 67 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Groot 65B A6B 261023 | 70Great quality (A) · 70 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 85/100 (quality, fit, speed and memory use) sets the ranking. | 57.4 tok/s | Runs great | 49Fair quality (C) · 49 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen3.5 27B | 88Excellent quality (A+) · 88 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 85/100 (quality, fit, speed and memory use) sets the ranking. | 31.2 tok/s | Runs great | 65Good quality (B) · 65 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 3.6 tok/s | Slowfar over GPU memory limit |
| DeepSeek V2.5 1210 | 68Good quality (B) · 68 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 85/100 (quality, fit, speed and memory use) sets the ranking. | 36.2 tok/s | Runs great | 50Fair quality (C) · 50 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Ling 3.0 flash Sante | 68Good quality (B) · 68 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 85/100 (quality, fit, speed and memory use) sets the ranking. | 65.0 tok/s | Runs great | 47Fair quality (C) · 47 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Sarvam 105B | 74Great quality (A) · 74 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 85/100 (quality, fit, speed and memory use) sets the ranking. | 49.4 tok/s | Runs great | 52Fair quality (C) · 52 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Nemotron 3 Nano 30B-A3B | 77Great quality (A) · 77 of 100: how capable the model is at F16, whatever it runs on. Match score 85/100 (quality, fit, speed and memory use) sets the ranking. | 46.3 tok/s | Runs great | 53Fair quality (C) · 53 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen3 30B-A3B Instruct 2507 | 77Great quality (A) · 77 of 100: how capable the model is at F16, whatever it runs on. Match score 85/100 (quality, fit, speed and memory use) sets the ranking. | 48.5 tok/s | Runs great | 53Fair quality (C) · 53 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| DiffusionGemma 26B-A4B IT | 78Great quality (A) · 78 of 100: how capable the model is at F16, whatever it runs on. Match score 85/100 (quality, fit, speed and memory use) sets the ranking. | 41.5 tok/s | Runs great | 54Fair quality (C) · 54 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 23.0 tok/s | Slowfar over GPU memory limit |
| Muse Glimmer 30B | 87Excellent quality (A+) · 87 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 85/100 (quality, fit, speed and memory use) sets the ranking. | 29.1 tok/s | Runs well | 64Good quality (B) · 64 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen3-VL 30B-A3B Thinking | 76Great quality (A) · 76 of 100: how capable the model is at F16, whatever it runs on. Match score 84/100 (quality, fit, speed and memory use) sets the ranking. | 48.5 tok/s | Runs great | 53Fair quality (C) · 53 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| DeepSeek V3 (671B) | 63Good quality (B) · 63 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 84/100 (quality, fit, speed and memory use) sets the ranking. | 27.5 tok/s | Runs welltight fit | 52Fair quality (C) · 52 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| DeepSeek V2.5 | 66Good quality (B) · 66 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 84/100 (quality, fit, speed and memory use) sets the ranking. | 36.2 tok/s | Runs great | 49Fair quality (C) · 49 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| TMax 27B | 85Excellent quality (A+) · 85 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 84/100 (quality, fit, speed and memory use) sets the ranking. | 31.6 tok/s | Runs great | 63Good quality (B) · 63 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 3.7 tok/s | Slowfar over GPU memory limit |
| Gemma 4 31B | 86Excellent quality (A+) · 86 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 84/100 (quality, fit, speed and memory use) sets the ranking. | 27.7 tok/s | Runs well | 63Good quality (B) · 63 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Gemma 4 12B | 82Great quality (A) · 82 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 84/100 (quality, fit, speed and memory use) sets the ranking. | 50.2 tok/s | Runs great | 80Great quality (A) · 80 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 85/100 (quality, fit, speed and memory use) sets the ranking. | 5.0 tok/s | Slowtight fit |
| Qwen3-VL 30B-A3B Instruct | 74Great quality (A) · 74 of 100: how capable the model is at F16, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | 48.5 tok/s | Runs great | 51Fair quality (C) · 51 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen3 30B-A3B | 74Great quality (A) · 74 of 100: how capable the model is at F16, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | 48.5 tok/s | Runs great | 51Fair quality (C) · 51 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Fara 1.5 27B | 83Great quality (A) · 83 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | 31.6 tok/s | Runs great | 61Good quality (B) · 61 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 3.7 tok/s | Slowfar over GPU memory limit |
| Hunyuan A13B (80B-A13B) | 73Great quality (A) · 73 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | 39.6 tok/s | Runs great | 51Fair quality (C) · 51 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Hunyuan Hy-Memory 30B-A3B | 85Excellent quality (A+) · 85 of 100: how capable the model is at F16, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 59Good quality (B) · 59 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Ring Flash 2.0 (100B-A6B) | 78Great quality (A) · 78 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | 56.7 tok/s | Runs great | 54Fair quality (C) · 54 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| K2 Horizon 7B | 80Great quality (A) · 80 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | 49.0 tok/s | Runs great | 67Good quality (B) · 67 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | 10.4 tok/s | Runs well |
| Nemotron 3.5 Lightning 30B-A3B | 72Great quality (A) · 72 of 100: how capable the model is at F16, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 50Fair quality (C) · 50 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| LFM2.5 24B-A2B | 74Great quality (A) · 74 of 100: how capable the model is at F16, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | 60.8 tok/s | Runs great | 51Fair quality (C) · 51 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 35.7 tok/s | Runs wellover GPU memory limit |
| UI-Mate 27B | 81Great quality (A) · 81 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | 31.6 tok/s | Runs great | 60Good quality (B) · 60 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | 3.7 tok/s | Slowfar over GPU memory limit |
| Qwen3-Omni 30B-A3B Instruct | 70Great quality (A) · 70 of 100: how capable the model is at F16, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | 48.5 tok/s | Runs great | 49Fair quality (C) · 49 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| SenseNova Loop 9B Preview | 78Great quality (A) · 78 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | 45.6 tok/s | Runs great | 54Fair quality (C) · 54 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 12.3 tok/s | Runs well |
| Llama 3.1 Nemotron Ultra 253B v1 | 81Great quality (A) · 81 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | 3.4 tok/s | Slow | 60Good quality (B) · 60 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Llama 4 Scout 17B-16E | 69Good quality (B) · 69 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | 36.7 tok/s | Runs great | 50Fair quality (C) · 50 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Sarvam 30B | 70Great quality (A) · 70 of 100: how capable the model is at F16, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | 61.8 tok/s | Runs great | 49Fair quality (C) · 49 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Ornith 1.5 9B | 77Great quality (A) · 77 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | 45.7 tok/s | Runs great | 53Fair quality (C) · 53 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 12.3 tok/s | Runs well |
| Qwen3.5 9B | 76Great quality (A) · 76 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | 45.7 tok/s | Runs great | 53Fair quality (C) · 53 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 12.3 tok/s | Runs well |
| North Mini Code 1.0 (30B-A3B) | 80Great quality (A) · 80 of 100: how capable the model is at F16, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 56Good quality (B) · 56 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| DeepSeek Coder V2 (236B) | 70Great quality (A) · 70 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | 36.2 tok/s | Runs great | 51Fair quality (C) · 51 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Snowflake Arctic Instruct | 47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | 43.5 tok/s | Runs great | 35Basic quality (D) · 35 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 4/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| DeepSeek V2 Chat | 58Good quality (B) · 58 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | 36.2 tok/s | Runs great | 43Fair quality (C) · 43 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 5/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Ornith 1.0 9B | 75Great quality (A) · 75 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | 45.7 tok/s | Runs great | 52Fair quality (C) · 52 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 12.3 tok/s | Runs well |
| gpt-oss 20B | 70Great quality (A) · 70 of 100: how capable the model is at F16, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | 45.3 tok/s | Runs great | 70Great quality (A) · 70 of 100: how capable the model is at MXFP4, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| TMax 9B | 74Great quality (A) · 74 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | 49.2 tok/s | Runs great | 62Good quality (B) · 62 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | 10.4 tok/s | Runs well |
| Qwen3 30B-A3B Thinking 2507 | 78Great quality (A) · 78 of 100: how capable the model is at F16, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | 48.5 tok/s | Runs great | 54Fair quality (C) · 54 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen3 Coder 30B-A3B | 78Great quality (A) · 78 of 100: how capable the model is at F16, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | 48.5 tok/s | Runs great | 54Fair quality (C) · 54 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| DeepSeek Coder V2 Instruct (0614) | 68Good quality (B) · 68 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | 36.2 tok/s | Runs great | 50Fair quality (C) · 50 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| EXAONE 4.5 33B | 77Great quality (A) · 77 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | 25.2 tok/s | Runs well | 57Good quality (B) · 57 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Apriel 1.5 15B Thinker | 71Great quality (A) · 71 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | 40.4 tok/s | Runs great | 60Good quality (B) · 60 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | 6.3 tok/s | Slowtight fit |
| Trinity Mini (26B-A3B) | 66Good quality (B) · 66 of 100: how capable the model is at F16, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 46Fair quality (C) · 46 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 29.9 tok/s | Slowfar over GPU memory limit |
| Devstral Small 2 (24B) | 72Great quality (A) · 72 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | 36.1 tok/s | Runs great | 53Fair quality (C) · 53 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | 4.2 tok/s | Slowfar over GPU memory limit |
| Nemotron 3.5 Bolt E4B | 70Great quality (A) · 70 of 100: how capable the model is at F16, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | 45.1 tok/s | Runs great | 66Good quality (B) · 66 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | 16.0 tok/s | Runs well |
| Seed-OSS 36B Instruct | 76Great quality (A) · 76 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | 23.9 tok/s | Runs well | 56Good quality (B) · 56 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Axion 1.0 35B | 62Good quality (B) · 62 of 100: how capable the model is at F16, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 43Fair quality (C) · 43 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 5/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Phi-4 Reasoning Vision 15B | 69Good quality (B) · 69 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | 39.7 tok/s | Runs great | 58Good quality (B) · 58 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | 6.2 tok/s | Slowtight fit |
| Hermes 4.3 36B | 75Great quality (A) · 75 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | 23.9 tok/s | Runs well | 55Good quality (B) · 55 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen3-VL 32B Instruct | 74Great quality (A) · 74 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | 26.0 tok/s | Runs well | 54Fair quality (C) · 54 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Granite 4.2 30B | 72Great quality (A) · 72 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | 29.6 tok/s | Runs well | 53Fair quality (C) · 53 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Gemma 4 E4B | 68Good quality (B) · 68 of 100: how capable the model is at F16, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | 45.1 tok/s | Runs great | 64Good quality (B) · 64 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | 16.0 tok/s | Runs well |
| ERNIE 4.5 21B-A3B | 64Good quality (B) · 64 of 100: how capable the model is at F16, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 44Fair quality (C) · 44 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | 35.2 tok/s | Runs welltight fit |
| Jamba 1.5 Large | 58Good quality (B) · 58 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | 11.5 tok/s | Runs well | 49Fair quality (C) · 49 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Granite 4.2 8B | 68Good quality (B) · 68 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | 50.1 tok/s | Runs great | 57Good quality (B) · 57 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | 10.6 tok/s | Runs well |
| Nanbeige4.2 3B | 68Good quality (B) · 68 of 100: how capable the model is at F16, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | 48.7 tok/s | Runs great | 64Good quality (B) · 64 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | 17.3 tok/s | Runs well |
| Nemotron Labs Dory 1.0 4B | 68Good quality (B) · 68 of 100: how capable the model is at F16, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | 51.0 tok/s | Runs great | 64Good quality (B) · 64 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | 18.1 tok/s | Runs well |
| Qwen3 14B | 67Good quality (B) · 67 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | 40.6 tok/s | Runs great | 57Good quality (B) · 57 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 6.3 tok/s | Slowtight fit |
| Llama 3.3 Nemotron Super 49B v1.5 | 76Great quality (A) · 76 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | 17.4 tok/s | Runs well | 56Good quality (B) · 56 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen3 32B | 72Great quality (A) · 72 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | 26.4 tok/s | Runs well | 53Fair quality (C) · 53 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| NorthCUA Mini | 59Good quality (B) · 59 of 100: how capable the model is at F16, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 41Fair quality (C) · 41 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 5/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| MiMo V2.6 Distill Qwen 9B | 78Great quality (A) · 78 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | 46.8 tok/s | Runs great | 54Fair quality (C) · 54 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 12.6 tok/s | Runs well |
| InternVL3.5 38B | 72Great quality (A) · 72 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | 22.6 tok/s | Runs well | 53Fair quality (C) · 53 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Devstral Small 1.1 (24B) | 66Good quality (B) · 66 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | 36.7 tok/s | Runs great | 49Fair quality (C) · 49 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | 4.3 tok/s | Slowfar over GPU memory limit |
| Magistral Small 2507 | 66Good quality (B) · 66 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | 36.7 tok/s | Runs great | 49Fair quality (C) · 49 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | 4.3 tok/s | Slowfar over GPU memory limit |
| Spark-X2.6 4B | 66Good quality (B) · 66 of 100: how capable the model is at F16, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | 49.4 tok/s | Runs great | 62Good quality (B) · 62 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | 17.6 tok/s | Runs well |
| Hermes 4 14B | 65Good quality (B) · 65 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | 40.7 tok/s | Runs great | 55Good quality (B) · 55 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 6.3 tok/s | Slowtight fit |
| Gemma 3 27B IT | 68Good quality (B) · 68 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | 31.6 tok/s | Runs great | 50Fair quality (C) · 50 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | 3.7 tok/s | Slowfar over GPU memory limit |
| GLM-4 32B 0414 | 70Great quality (A) · 70 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | 26.6 tok/s | Runs well | 51Fair quality (C) · 51 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Granite 4.0 H Small (32B-A9B) | 62Good quality (B) · 62 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | 41.3 tok/s | Runs great | 43Fair quality (C) · 43 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 5/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Granite 4.1 30B | 68Good quality (B) · 68 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | 30.0 tok/s | Runs well | 50Fair quality (C) · 50 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| MedGemma 27B Text IT | 66Good quality (B) · 66 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | 32.1 tok/s | Runs great | 49Fair quality (C) · 49 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | 3.7 tok/s | Slowfar over GPU memory limit |
| Magistral Small 2506 | 64Good quality (B) · 64 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | 36.7 tok/s | Runs great | 47Fair quality (C) · 47 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | 4.3 tok/s | Slowfar over GPU memory limit |
| Phi-4 (14B) | 63Good quality (B) · 63 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | 40.9 tok/s | Runs great | 53Fair quality (C) · 53 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 6.3 tok/s | Slowtight fit |
| EXAONE 4.0 32B | 68Good quality (B) · 68 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | 27.1 tok/s | Runs well | 50Fair quality (C) · 50 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| InternVL3 14B | 63Good quality (B) · 63 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | 39.7 tok/s | Runs great | 53Fair quality (C) · 53 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 6.2 tok/s | Slowtight fit |
| Olmo 3.1 32B Instruct | 68Good quality (B) · 68 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | 26.9 tok/s | Runs well | 50Fair quality (C) · 50 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Ministral 3 14B | 63Good quality (B) · 63 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | 43.1 tok/s | Runs great | 53Fair quality (C) · 53 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 6.7 tok/s | Slowtight fit |
| Mistral Small 3.2 24B | 64Good quality (B) · 64 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | 36.1 tok/s | Runs great | 47Fair quality (C) · 47 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | 4.2 tok/s | Slowfar over GPU memory limit |
| ERNIE 4.5 21B-A3B Thinking | 70Great quality (A) · 70 of 100: how capable the model is at F16, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 49Fair quality (C) · 49 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 35.2 tok/s | Runs welltight fit |
| Kimi-VL A3B Thinking 2506 | 60Good quality (B) · 60 of 100: how capable the model is at F16, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 50Fair quality (C) · 50 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | 28.2 tok/s | Runs welltight fit |
| Maple Preview (20B-A1B) | 70Great quality (A) · 70 of 100: how capable the model is at F16, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 85.4 tok/s | Runs great | 49Fair quality (C) · 49 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | 63.7 tok/s | Runs welltight fit |
| Qwen3-VL 8B Instruct | 62Good quality (B) · 62 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 50.3 tok/s | Runs great | 52Fair quality (C) · 52 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | 10.6 tok/s | Runs well |
| Gemma 4 E2B | 62Good quality (B) · 62 of 100: how capable the model is at F16, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 88.3 tok/s | Runs great | 58Good quality (B) · 58 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | 31.4 tok/s | Runs great |
| InternVL3.5 8B | 62Good quality (B) · 62 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 51.7 tok/s | Runs great | 52Fair quality (C) · 52 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | 10.9 tok/s | Runs well |
| Qwen3.5 4B | 62Good quality (B) · 62 of 100: how capable the model is at F16, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 43.6 tok/s | Runs great | 58Good quality (B) · 58 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 15.5 tok/s | Runs well |
| Baichuan M2 32B | 66Good quality (B) · 66 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 26.4 tok/s | Runs well | 49Fair quality (C) · 49 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen2.5 14B Instruct | 61Good quality (B) · 61 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 40.9 tok/s | Runs great | 52Fair quality (C) · 52 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 6.3 tok/s | Slowtight fit |
| Qwen2.5 32B Instruct | 66Good quality (B) · 66 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 26.4 tok/s | Runs well | 49Fair quality (C) · 49 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Mistral Small 3.1 24B | 62Good quality (B) · 62 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 36.1 tok/s | Runs great | 46Fair quality (C) · 46 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 49/100 (quality, fit, speed and memory use) sets the ranking. | 4.2 tok/s | Slowfar over GPU memory limit |
| Terminal RSI 27B | 64Good quality (B) · 64 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 31.2 tok/s | Runs great | 47Fair quality (C) · 47 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 49/100 (quality, fit, speed and memory use) sets the ranking. | 3.6 tok/s | Slowfar over GPU memory limit |
| Hemmingway 1.5 | 63Good quality (B) · 63 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 32.2 tok/s | Runs great | 46Fair quality (C) · 46 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 49/100 (quality, fit, speed and memory use) sets the ranking. | 3.7 tok/s | Slowfar over GPU memory limit |
| LFM2 24B-A2B | 56Good quality (B) · 56 of 100: how capable the model is at F16, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 63.7 tok/s | Runs great | 39Basic quality (D) · 39 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 37.7 tok/s | Runs wellover GPU memory limit |
| Groot 19B A3B 261023 | 57Good quality (B) · 57 of 100: how capable the model is at F16, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 39Basic quality (D) · 39 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | 35.2 tok/s | Runs welltight fit |
| TMax 4B | 61Good quality (B) · 61 of 100: how capable the model is at F16, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 48.2 tok/s | Runs great | 57Good quality (B) · 57 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 17.1 tok/s | Runs well |
| Mellum2.1 Thinking (12B-A2.5B) | 70Great quality (A) · 70 of 100: how capable the model is at F16, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 59.9 tok/s | Runs great | 66Good quality (B) · 66 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | 26.3 tok/s | Runs well |
| Ling 3.0 Tiny (8B-A1.3B) | 60Good quality (B) · 60 of 100: how capable the model is at F16, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 92.8 tok/s | Runs great | 58Good quality (B) · 58 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 86/100 (quality, fit, speed and memory use) sets the ranking. | 39.5 tok/s | Runs great |
| Phi-3.5 MoE (42B-A6.6B) | 56Good quality (B) · 56 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 53.2 tok/s | Runs great | 39Basic quality (D) · 39 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 5/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Phi-4 Reasoning Plus (14B) | 71Great quality (A) · 71 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 41.0 tok/s | Runs great | 60Good quality (B) · 60 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 6.4 tok/s | Slowtight fit |
| InternVL3 38B | 66Good quality (B) · 66 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 22.6 tok/s | Runs well | 49Fair quality (C) · 49 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Hermes 4 70B | 70Great quality (A) · 70 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 12.3 tok/s | Runs well | 51Fair quality (C) · 51 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Llama 3.3 70B Instruct | 70Great quality (A) · 70 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 12.3 tok/s | Runs well | 51Fair quality (C) · 51 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Mistral Small 3 24B (2501) | 60Good quality (B) · 60 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 36.7 tok/s | Runs great | 44Fair quality (C) · 44 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 49/100 (quality, fit, speed and memory use) sets the ranking. | 4.3 tok/s | Slowfar over GPU memory limit |
| GLM-4.1V 9B Thinking | 60Good quality (B) · 60 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 42.8 tok/s | Runs great | 42Fair quality (C) · 42 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 11.5 tok/s | Runs well |
| Granite 4.1 8B | 60Good quality (B) · 60 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 50.1 tok/s | Runs great | 50Fair quality (C) · 50 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 10.6 tok/s | Runs well |
| MiniCPM-V 4.5 8B | 60Good quality (B) · 60 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 50.7 tok/s | Runs great | 50Fair quality (C) · 50 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 10.7 tok/s | Runs well |
| Qwen3 8B | 60Good quality (B) · 60 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 53.8 tok/s | Runs great | 50Fair quality (C) · 50 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 11.4 tok/s | Runs well |
| Nemotron Nano 9B v2 | 60Good quality (B) · 60 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 49.6 tok/s | Runs great | 50Fair quality (C) · 50 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 10.5 tok/s | Runs well |
| Agents-A1 4B | 60Good quality (B) · 60 of 100: how capable the model is at F16, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 44.7 tok/s | Runs great | 57Good quality (B) · 57 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 15.9 tok/s | Runs well |
| Qwen2.5 14B Instruct 1M | 59Good quality (B) · 59 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 40.7 tok/s | Runs great | 50Fair quality (C) · 50 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 6.3 tok/s | Slowtight fit |
| InternVL3 78B | 70Great quality (A) · 70 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 11.0 tok/s | Runs well | 51Fair quality (C) · 51 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen2.5-VL 32B Instruct | 64Good quality (B) · 64 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 25.9 tok/s | Runs well | 47Fair quality (C) · 47 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Gemma 3 12B IT | 59Good quality (B) · 59 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 49.2 tok/s | Runs great | 57Good quality (B) · 57 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 5.9 tok/s | Slowtight fit |
| MiniCPM5 2B | 60Good quality (B) · 60 of 100: how capable the model is at F16, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 80.6 tok/s | Runs great | 57Good quality (B) · 57 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 28.6 tok/s | Runs well |
| Qwen2.5 72B Instruct | 69Good quality (B) · 69 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 11.9 tok/s | Runs well | 51Fair quality (C) · 51 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen2.5-VL 72B Instruct | 69Good quality (B) · 69 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 11.8 tok/s | Runs well | 51Fair quality (C) · 51 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| UI-Mate 9B | 70Great quality (A) · 70 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 46.8 tok/s | Runs great | 49Fair quality (C) · 49 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 12.6 tok/s | Runs well |
| UI-Venus 2 9B | 70Great quality (A) · 70 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 46.8 tok/s | Runs great | 49Fair quality (C) · 49 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 12.6 tok/s | Runs well |
| Phi-4 Reasoning (14B) | 69Good quality (B) · 69 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 41.0 tok/s | Runs great | 58Good quality (B) · 58 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 6.4 tok/s | Slowtight fit |
| QwQ 32B | 74Great quality (A) · 74 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 26.4 tok/s | Runs well | 54Fair quality (C) · 54 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| DeepSeek R1 Distill Llama 70B | 68Good quality (B) · 68 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 12.3 tok/s | Runs well | 50Fair quality (C) · 50 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Tülu 3 70B | 68Good quality (B) · 68 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 12.3 tok/s | Runs well | 50Fair quality (C) · 50 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Dolphin Mistral 24B Venice Edition | 58Good quality (B) · 58 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 36.1 tok/s | Runs great | 43Fair quality (C) · 43 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | 4.2 tok/s | Slowfar over GPU memory limit |
| MiMo-VL 7B RL | 58Good quality (B) · 58 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 53.0 tok/s | Runs great | 48Fair quality (C) · 48 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 11.2 tok/s | Runs well |
| Fara 1.5 4B | 58Good quality (B) · 58 of 100: how capable the model is at F16, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 44.7 tok/s | Runs great | 55Good quality (B) · 55 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 15.9 tok/s | Runs well |
| Voxtral Small 24B | 58Good quality (B) · 58 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 35.7 tok/s | Runs great | 43Fair quality (C) · 43 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | 4.2 tok/s | Slowfar over GPU memory limit |
| Solar Pro Preview 22B | 57Good quality (B) · 57 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 39.1 tok/s | Runs great | 42Fair quality (C) · 42 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | 4.5 tok/s | Slowover GPU memory limit |
| Command A (111B) | 67Good quality (B) · 67 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 10.1 tok/s | Runs well | 56Good quality (B) · 56 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| OpenReasoning Nemotron 32B | 72Great quality (A) · 72 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 26.4 tok/s | Runs well | 53Fair quality (C) · 53 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Skywork OR1 32B | 72Great quality (A) · 72 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 26.4 tok/s | Runs well | 53Fair quality (C) · 53 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Olmo 3.1 32B Think | 72Great quality (A) · 72 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 26.9 tok/s | Runs well | 53Fair quality (C) · 53 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Gemma 2 27B IT | 58Good quality (B) · 58 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 31.8 tok/s | Runs great | 43Fair quality (C) · 43 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 5/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Magistral Small 1.2 (24B) | 68Good quality (B) · 68 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 36.1 tok/s | Runs great | 50Fair quality (C) · 50 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 46/100 (quality, fit, speed and memory use) sets the ranking. | 4.2 tok/s | Slowfar over GPU memory limit |
| Llama 3.1 70B Instruct | 66Good quality (B) · 66 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 12.3 tok/s | Runs well | 49Fair quality (C) · 49 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Dolphin 3.0 R1 Mistral 24B | 57Good quality (B) · 57 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 36.7 tok/s | Runs great | 42Fair quality (C) · 42 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 47/100 (quality, fit, speed and memory use) sets the ranking. | 4.3 tok/s | Slowfar over GPU memory limit |
| Mage-VL 4.7B | 56Good quality (B) · 56 of 100: how capable the model is at F16, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 42.9 tok/s | Runs great | 53Fair quality (C) · 53 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 15.2 tok/s | Runs well |
| Ministral 3 8B | 56Good quality (B) · 56 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 49.4 tok/s | Runs great | 47Fair quality (C) · 47 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 10.5 tok/s | Runs well |
| MiMo 7B RL | 56Good quality (B) · 56 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 56.3 tok/s | Runs great | 47Fair quality (C) · 47 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 11.9 tok/s | Runs well |
| InternVL3 8B | 56Good quality (B) · 56 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 55.5 tok/s | Runs great | 47Fair quality (C) · 47 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 11.7 tok/s | Runs well |
| Falcon-H1 34B Instruct | 60Good quality (B) · 60 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 25.7 tok/s | Runs well | 44Fair quality (C) · 44 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Nous Hermes 2 Mixtral 8x7B DPO | 51Fair quality (C) · 51 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 39.9 tok/s | Runs great | 36Basic quality (D) · 36 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 4/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| LFM3 1.7B | 56Good quality (B) · 56 of 100: how capable the model is at F16, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 117 tok/s | Runs great | 53Fair quality (C) · 53 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 41.7 tok/s | Runs great |
| Ling Mini 2.0 (16B-A1.4B) | 52Fair quality (C) · 52 of 100: how capable the model is at F16, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 88.7 tok/s | Runs great | 43Fair quality (C) · 43 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | 54.8 tok/s | Runs welltight fit |
| EXAONE Deep 32B | 70Great quality (A) · 70 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 27.1 tok/s | Runs well | 51Fair quality (C) · 51 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Apriel Nemotron 15B Thinker | 65Good quality (B) · 65 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 40.1 tok/s | Runs great | 55Good quality (B) · 55 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 6.2 tok/s | Slowtight fit |
| AceReason Nemotron 14B | 65Good quality (B) · 65 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 40.7 tok/s | Runs great | 55Good quality (B) · 55 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 6.3 tok/s | Slowtight fit |
| OpenReasoning Nemotron 14B | 65Good quality (B) · 65 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 40.7 tok/s | Runs great | 55Good quality (B) · 55 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 6.3 tok/s | Slowtight fit |
| Qwen2.5 Coder 14B Instruct | 65Good quality (B) · 65 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 40.7 tok/s | Runs great | 55Good quality (B) · 55 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 6.3 tok/s | Slowtight fit |
| Kimi Dev 72B | 76Great quality (A) · 76 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 11.9 tok/s | Runs well | 56Good quality (B) · 56 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Ministral 3 14B Reasoning | 65Good quality (B) · 65 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 43.1 tok/s | Runs great | 55Good quality (B) · 55 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 6.7 tok/s | Slowtight fit |
| Hermes 3 (Llama 3.1 70B) | 64Good quality (B) · 64 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 12.3 tok/s | Runs well | 47Fair quality (C) · 47 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| GLM-Z1 32B (0414) | 70Great quality (A) · 70 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 26.6 tok/s | Runs well | 51Fair quality (C) · 51 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| QVQ 72B Preview | 64Good quality (B) · 64 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 11.8 tok/s | Runs well | 47Fair quality (C) · 47 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Command R (08-2024) 32B | 58Good quality (B) · 58 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 26.8 tok/s | Runs well | 43Fair quality (C) · 43 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 5/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen3-VL 4B Instruct | 54Fair quality (C) · 54 of 100: how capable the model is at F16, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 45.7 tok/s | Runs great | 51Fair quality (C) · 51 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 16.2 tok/s | Runs well |
| Llama 3.1 Nemotron Nano 8B v1 | 54Fair quality (C) · 54 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 54.9 tok/s | Runs great | 45Fair quality (C) · 45 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 11.6 tok/s | Runs well |
| Qwen3 4B Instruct 2507 | 54Fair quality (C) · 54 of 100: how capable the model is at F16, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 50.5 tok/s | Runs great | 51Fair quality (C) · 51 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 17.9 tok/s | Runs well |
| MiniCPM4 8B | 54Fair quality (C) · 54 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 53.8 tok/s | Runs great | 45Fair quality (C) · 45 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 11.4 tok/s | Runs well |
| Jamba 1.5 Mini | 49Fair quality (C) · 49 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 42.4 tok/s | Runs great | 35Basic quality (D) · 35 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 4/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Inkling (975B-A41B) | 65Good quality (B) · 65 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 26.6 tok/s | Runs wellover GPU memory limit | 65Good quality (B) · 65 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Hermes 4 405B | 56Good quality (B) · 56 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 3.5 tok/s | Slow | 56Good quality (B) · 56 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Mixtral 8x7B Instruct | 49Fair quality (C) · 49 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 39.9 tok/s | Runs great | 35Basic quality (D) · 35 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 4/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Mixtral 8x22B Instruct | 52Fair quality (C) · 52 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 20.7 tok/s | Runs well | 38Basic quality (D) · 38 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 5/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Kimi-VL A3B Instruct | 50Fair quality (C) · 50 of 100: how capable the model is at F16, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 42Fair quality (C) · 42 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 28.2 tok/s | Runs welltight fit |
| LFM2.5 8B-A1B | 52Fair quality (C) · 52 of 100: how capable the model is at F16, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 85.0 tok/s | Runs great | 49Fair quality (C) · 49 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | 41.4 tok/s | Runs great |
| DeepCoder 14B Preview | 63Good quality (B) · 63 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 40.7 tok/s | Runs great | 53Fair quality (C) · 53 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 6.3 tok/s | Slowtight fit |
| Qwen2 57B-A14B Instruct | 48Fair quality (C) · 48 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 43.3 tok/s | Runs great | 35Basic quality (D) · 35 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 4/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| DeepSeek R1 Distill Qwen 32B | 68Good quality (B) · 68 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 26.4 tok/s | Runs well | 50Fair quality (C) · 50 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Llama 3.1 Nemotron 70B Instruct | 62Good quality (B) · 62 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 12.3 tok/s | Runs well | 46Fair quality (C) · 46 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Apertus v1.5 70B | 62Good quality (B) · 62 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 12.0 tok/s | Runs well | 46Fair quality (C) · 46 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Aya Expanse 32B | 57Good quality (B) · 57 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 26.8 tok/s | Runs well | 42Fair quality (C) · 42 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 5/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| OLMo 2 32B Instruct | 57Good quality (B) · 57 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 26.9 tok/s | Runs well | 42Fair quality (C) · 42 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 5/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Falcon 3 10B Instruct | 52Fair quality (C) · 52 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 42.7 tok/s | Runs great | 36Basic quality (D) · 36 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 11.5 tok/s | Runs well |
| Aya Vision 32B | 57Good quality (B) · 57 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 26.1 tok/s | Runs well | 42Fair quality (C) · 42 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 5/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Reka Flash 3.1 (21B) | 62Good quality (B) · 62 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 41.4 tok/s | Runs great | 46Fair quality (C) · 46 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | 4.8 tok/s | Slowover GPU memory limit |
| GLM-4 9B 0414 | 52Fair quality (C) · 52 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 46.9 tok/s | Runs great | 36Basic quality (D) · 36 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 12.6 tok/s | Runs well |
| NorthCUA Nano | 52Fair quality (C) · 52 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 46.8 tok/s | Runs great | 36Basic quality (D) · 36 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 12.6 tok/s | Runs well |
| MiniCPM-o 2.6 8B | 52Fair quality (C) · 52 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 50.8 tok/s | Runs great | 43Fair quality (C) · 43 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 10.8 tok/s | Runs well |
| Qwen2.5-VL 7B Instruct | 52Fair quality (C) · 52 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 53.2 tok/s | Runs great | 43Fair quality (C) · 43 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 11.3 tok/s | Runs well |
| Hunyuan 7B Instruct | 52Fair quality (C) · 52 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 58.8 tok/s | Runs great | 43Fair quality (C) · 43 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 12.4 tok/s | Runs well |
| Qwen2.5 7B Instruct | 52Fair quality (C) · 52 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 57.8 tok/s | Runs great | 43Fair quality (C) · 43 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 12.2 tok/s | Runs well |
| Granite 4.2 3B | 52Fair quality (C) · 52 of 100: how capable the model is at F16, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 55.5 tok/s | Runs great | 49Fair quality (C) · 49 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 19.7 tok/s | Runs well |
| Mistral Nemo 12B Instruct | 51Fair quality (C) · 51 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 49.2 tok/s | Runs great | 49Fair quality (C) · 49 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 5.9 tok/s | Slowtight fit |
| Mistral Small 22B (2409) | 51Fair quality (C) · 51 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 38.9 tok/s | Runs great | 37Basic quality (D) · 37 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 46/100 (quality, fit, speed and memory use) sets the ranking. | 4.5 tok/s | Slowfar over GPU memory limit |
| Devstral Small 2505 | 62Good quality (B) · 62 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 36.7 tok/s | Runs great | 46Fair quality (C) · 46 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 44/100 (quality, fit, speed and memory use) sets the ranking. | 4.3 tok/s | Slowfar over GPU memory limit |
| DeepSeek R1 0528 Qwen3 8B | 62Good quality (B) · 62 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 53.8 tok/s | Runs great | 52Fair quality (C) · 52 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 11.4 tok/s | Runs well |
| OpenHands LM 32B | 66Good quality (B) · 66 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 26.4 tok/s | Runs well | 49Fair quality (C) · 49 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| DeepSeek R1 Distill Qwen 14B | 61Good quality (B) · 61 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 40.7 tok/s | Runs great | 52Fair quality (C) · 52 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 6.3 tok/s | Slowtight fit |
| JanusCoder 14B | 61Good quality (B) · 61 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 40.7 tok/s | Runs great | 52Fair quality (C) · 52 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 6.3 tok/s | Slowtight fit |
| Tülu 3 405B | 53Fair quality (C) · 53 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 3.5 tok/s | Slow | 53Fair quality (C) · 53 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen2 72B Instruct | 60Good quality (B) · 60 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 11.9 tok/s | Runs well | 44Fair quality (C) · 44 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen2-VL 72B Instruct | 60Good quality (B) · 60 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 11.8 tok/s | Runs well | 44Fair quality (C) · 44 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Gemma 2 9B IT | 50Fair quality (C) · 50 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 47.7 tok/s | Runs great | 42Fair quality (C) · 42 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 10.1 tok/s | Runs well |
| InternLM3 8B Instruct | 50Fair quality (C) · 50 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 50.1 tok/s | Runs great | 42Fair quality (C) · 42 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 10.6 tok/s | Runs well |
| InternVL2.5 8B | 50Fair quality (C) · 50 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 54.5 tok/s | Runs great | 42Fair quality (C) · 42 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 11.5 tok/s | Runs well |
| Foundation-Sec 1.1 8B Instruct | 50Fair quality (C) · 50 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 54.9 tok/s | Runs great | 42Fair quality (C) · 42 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 11.6 tok/s | Runs well |
| Qwen3 4B | 50Fair quality (C) · 50 of 100: how capable the model is at F16, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 50.5 tok/s | Runs great | 47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 17.9 tok/s | Runs well |
| Olmo 3 7B Instruct | 50Fair quality (C) · 50 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 60.4 tok/s | Runs great | 42Fair quality (C) · 42 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 12.8 tok/s | Runs well |
| GLiDE 9B | 50Fair quality (C) · 50 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 55.5 tok/s | Runs great | 42Fair quality (C) · 42 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 11.7 tok/s | Runs well |
| Qwen2.5 7B Instruct 1M | 50Fair quality (C) · 50 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 57.8 tok/s | Runs great | 42Fair quality (C) · 42 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 12.2 tok/s | Runs well |
| Falcon-H1 7B Instruct | 50Fair quality (C) · 50 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 58.1 tok/s | Runs great | 42Fair quality (C) · 42 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 12.3 tok/s | Runs well |
| Nemotron 3 Nano 4B | 50Fair quality (C) · 50 of 100: how capable the model is at F16, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 51.2 tok/s | Runs great | 47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 18.2 tok/s | Runs well |
| Pixtral 12B (2409) | 49Fair quality (C) · 49 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 48.4 tok/s | Runs great | 47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 5.8 tok/s | Slowtight fit |
| LFM3-VL 2B | 50Fair quality (C) · 50 of 100: how capable the model is at F16, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 106 tok/s | Runs great | 47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 37.6 tok/s | Runs great |
| Mage-VL-Cowork 4.7B | 60Good quality (B) · 60 of 100: how capable the model is at F16, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 42.9 tok/s | Runs great | 57Good quality (B) · 57 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 15.2 tok/s | Runs well |
| Moonlight 16B-A3B Instruct | 46Fair quality (C) · 46 of 100: how capable the model is at F16, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 52.2 tok/s | Runs great | 38Basic quality (D) · 38 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 28.2 tok/s | Runs welltight fit |
| SenseNova 6.8 Code2Video | 62Good quality (B) · 62 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 31.6 tok/s | Runs great | 46Fair quality (C) · 46 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 44/100 (quality, fit, speed and memory use) sets the ranking. | 3.7 tok/s | Slowfar over GPU memory limit |
| Command R+ (08-2024) 104B | 58Good quality (B) · 58 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 10.8 tok/s | Runs well | 49Fair quality (C) · 49 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| GLM-4 9B Chat 1M | 48Fair quality (C) · 48 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 46.5 tok/s | Runs great | 33Basic quality (D) · 33 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 12.5 tok/s | Runs well |
| Codestral 22B | 58Good quality (B) · 58 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 38.9 tok/s | Runs great | 43Fair quality (C) · 43 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 43/100 (quality, fit, speed and memory use) sets the ranking. | 4.5 tok/s | Slowfar over GPU memory limit |
| GLM-4 9B Chat (2024) | 48Fair quality (C) · 48 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 46.9 tok/s | Runs great | 33Basic quality (D) · 33 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 12.6 tok/s | Runs well |
| InternLM2.5 20B Chat | 47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 43.6 tok/s | Runs great | 35Basic quality (D) · 35 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | 5.1 tok/s | Slowover GPU memory limit |
| Hermes 3 (Llama 3.1 8B) | 48Fair quality (C) · 48 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 54.9 tok/s | Runs great | 40Fair quality (C) · 40 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 11.6 tok/s | Runs well |
| Llama 3.1 8B Instruct | 48Fair quality (C) · 48 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 54.9 tok/s | Runs great | 40Fair quality (C) · 40 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 11.6 tok/s | Runs well |
| Qwen2.5 Coder 32B Instruct | 63Good quality (B) · 63 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 26.4 tok/s | Runs well | 46Fair quality (C) · 46 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| MiniCPM-V 2.6 8B | 48Fair quality (C) · 48 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 54.4 tok/s | Runs great | 40Fair quality (C) · 40 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 11.5 tok/s | Runs well |
| Llama 3.1 405B Instruct | 50Fair quality (C) · 50 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 3.5 tok/s | Slow | 50Fair quality (C) · 50 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Command A Reasoning (111B) | 68Good quality (B) · 68 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 10.1 tok/s | Runs well | 57Good quality (B) · 57 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Command R 35B (v01) | 51Fair quality (C) · 51 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 24.7 tok/s | Runs well | 37Basic quality (D) · 37 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 5/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| QwQ 32B Preview | 62Good quality (B) · 62 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 26.4 tok/s | Runs well | 46Fair quality (C) · 46 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| GLM-Z1 9B (0414) | 58Good quality (B) · 58 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 46.9 tok/s | Runs great | 40Fair quality (C) · 40 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 12.6 tok/s | Runs well |
| Apertus 70B Instruct | 57Good quality (B) · 57 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 12.3 tok/s | Runs well | 42Fair quality (C) · 42 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 5/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Seed-Coder 8B Reasoning | 58Good quality (B) · 58 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 53.4 tok/s | Runs great | 48Fair quality (C) · 48 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 11.3 tok/s | Runs well |
| Grok-1 (314B) | 42Fair quality (C) · 42 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 12.5 tok/s | Runs well | 35Basic quality (D) · 35 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 4/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Gemma 3n E4B | 46Fair quality (C) · 46 of 100: how capable the model is at F16, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 50.8 tok/s | Runs great | 43Fair quality (C) · 43 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 18.0 tok/s | Runs well |
| Mistral Medium 3.5 (128B) | 56Good quality (B) · 56 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 11.1 tok/s | Runs well | 56Good quality (B) · 56 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen2.5-Omni 7B | 46Fair quality (C) · 46 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 41.1 tok/s | Runs great | 32Basic quality (D) · 32 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 11.0 tok/s | Runs well |
| Mistral Large 2.1 (2411) | 56Good quality (B) · 56 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 11.6 tok/s | Runs well | 56Good quality (B) · 56 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Apertus v1.5 8B | 46Fair quality (C) · 46 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 49.5 tok/s | Runs great | 38Basic quality (D) · 38 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 10.5 tok/s | Runs well |
| Granite 3.2 8B Instruct | 46Fair quality (C) · 46 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 53.9 tok/s | Runs great | 38Basic quality (D) · 38 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 11.4 tok/s | Runs well |
| Granite 3.3 8B | 46Fair quality (C) · 46 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 53.9 tok/s | Runs great | 38Basic quality (D) · 38 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 11.4 tok/s | Runs well |
| Ministral 8B (2410) | 46Fair quality (C) · 46 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 55.0 tok/s | Runs great | 38Basic quality (D) · 38 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 11.6 tok/s | Runs well |
| Phi-3 Medium 128K Instruct | 45Fair quality (C) · 45 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 43.0 tok/s | Runs great | 38Basic quality (D) · 38 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 6.7 tok/s | Slowtight fit |
| DeepHermes 3 Llama 3 8B Preview | 46Fair quality (C) · 46 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 54.9 tok/s | Runs great | 38Basic quality (D) · 38 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 11.6 tok/s | Runs well |
| Dolphin 3.0 Llama 3.1 8B | 46Fair quality (C) · 46 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 54.9 tok/s | Runs great | 38Basic quality (D) · 38 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 11.6 tok/s | Runs well |
| Tülu 3 8B | 46Fair quality (C) · 46 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 54.9 tok/s | Runs great | 38Basic quality (D) · 38 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 11.6 tok/s | Runs well |
| EXAONE 3.5 7.8B | 46Fair quality (C) · 46 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 56.4 tok/s | Runs great | 38Basic quality (D) · 38 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 11.9 tok/s | Runs well |
| Falcon 3 7B Instruct | 46Fair quality (C) · 46 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 59.1 tok/s | Runs great | 38Basic quality (D) · 38 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 12.5 tok/s | Runs well |
| Phi-3 Medium (14B) | 45Fair quality (C) · 45 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 43.0 tok/s | Runs great | 38Basic quality (D) · 38 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 6.7 tok/s | Slowtight fit |
| Llama 3.2 90B Vision Instruct | 55Good quality (B) · 55 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 12.6 tok/s | Runs well | 46Fair quality (C) · 46 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Granite 4.1 3B | 46Fair quality (C) · 46 of 100: how capable the model is at F16, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 59.7 tok/s | Runs great | 43Fair quality (C) · 43 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 21.2 tok/s | Runs well |
| Kimi K2.6 (1T) | 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 33.2 tok/s | Slowfar over GPU memory limit | 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Llama 3 70B Instruct | 55Good quality (B) · 55 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 12.3 tok/s | Runs well | 40Fair quality (C) · 40 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 5/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Seed-Coder 8B Instruct | 56Good quality (B) · 56 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 53.4 tok/s | Runs great | 47Fair quality (C) · 47 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 11.3 tok/s | Runs well |
| Gemma 3 4B IT | 45Fair quality (C) · 45 of 100: how capable the model is at F16, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 47.2 tok/s | Runs great | 42Fair quality (C) · 42 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 16.8 tok/s | Runs well |
| Mistral Large 2 (123B) | 54Fair quality (C) · 54 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 11.6 tok/s | Runs well | 54Fair quality (C) · 54 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| EXAONE Deep 7.8B | 56Good quality (B) · 56 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 56.4 tok/s | Runs great | 47Fair quality (C) · 47 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 11.9 tok/s | Runs well |
| Pixtral Large 124B (2411) | 54Fair quality (C) · 54 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 11.5 tok/s | Runs well | 54Fair quality (C) · 54 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen3 4B Thinking 2507 | 56Good quality (B) · 56 of 100: how capable the model is at F16, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 50.5 tok/s | Runs great | 53Fair quality (C) · 53 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 17.9 tok/s | Runs well |
| Falcon-H1R 7B | 56Good quality (B) · 56 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 58.1 tok/s | Runs great | 47Fair quality (C) · 47 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 12.3 tok/s | Runs well |
| Granite 4.0 H Tiny (7B-A1B) | 44Fair quality (C) · 44 of 100: how capable the model is at F16, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 108 tok/s | Runs great | 43Fair quality (C) · 43 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | 42.4 tok/s | Runs great |
| LLM4Decompile 22B V2 | 55Good quality (B) · 55 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 38.9 tok/s | Runs great | 40Fair quality (C) · 40 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 42/100 (quality, fit, speed and memory use) sets the ranking. | 4.5 tok/s | Slowfar over GPU memory limit |
| Kimi K2.5 (1T) | 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 33.2 tok/s | Slowfar over GPU memory limit | 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| OLMo 2 13B Instruct | 43Fair quality (C) · 43 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 43.8 tok/s | Runs great | 30Basic quality (D) · 30 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 8.6 tok/s | Slowtight fit |
| Llama 3.1 Nemotron Nano 4B v1.1 | 44Fair quality (C) · 44 of 100: how capable the model is at F16, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 45.0 tok/s | Runs great | 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 16.0 tok/s | Runs well |
| Granite 3.1 8B Instruct | 44Fair quality (C) · 44 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 53.9 tok/s | Runs great | 36Basic quality (D) · 36 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 11.4 tok/s | Runs well |
| InternVL2 8B | 44Fair quality (C) · 44 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 54.5 tok/s | Runs great | 36Basic quality (D) · 36 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 11.5 tok/s | Runs well |
| Dolphin 2.9.4 Llama 3.1 8B | 44Fair quality (C) · 44 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 54.9 tok/s | Runs great | 36Basic quality (D) · 36 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 11.6 tok/s | Runs well |
| Hermes 2 Pro Llama 3 8B | 44Fair quality (C) · 44 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 54.9 tok/s | Runs great | 36Basic quality (D) · 36 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 11.6 tok/s | Runs well |
| OpenChat 3.6 8B | 44Fair quality (C) · 44 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 54.9 tok/s | Runs great | 36Basic quality (D) · 36 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 11.6 tok/s | Runs well |
| Aya Vision 8B | 44Fair quality (C) · 44 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 51.1 tok/s | Runs great | 36Basic quality (D) · 36 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 10.8 tok/s | Runs well |
| Qwen2-VL 7B Instruct | 44Fair quality (C) · 44 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 53.2 tok/s | Runs great | 36Basic quality (D) · 36 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 11.3 tok/s | Runs well |
| Phi-3 Small 8K Instruct | 44Fair quality (C) · 44 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 59.6 tok/s | Runs great | 36Basic quality (D) · 36 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 12.6 tok/s | Runs well |
| Command R7B | 44Fair quality (C) · 44 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 54.9 tok/s | Runs great | 36Basic quality (D) · 36 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 11.6 tok/s | Runs well |
| MedGemma 4B IT | 44Fair quality (C) · 44 of 100: how capable the model is at F16, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 47.2 tok/s | Runs great | 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 16.8 tok/s | Runs well |
| Phi-4 Mini (3.8B) | 44Fair quality (C) · 44 of 100: how capable the model is at F16, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 52.9 tok/s | Runs great | 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 18.8 tok/s | Runs well |
| Ministral 3 3B | 44Fair quality (C) · 44 of 100: how capable the model is at F16, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 52.8 tok/s | Runs great | 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 18.7 tok/s | Runs well |
| Qwen2 7B Instruct | 44Fair quality (C) · 44 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 57.8 tok/s | Runs great | 36Basic quality (D) · 36 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 12.2 tok/s | Runs well |
| DeepSeek Coder V2 Lite (16B) | 52Fair quality (C) · 52 of 100: how capable the model is at F16, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 61.8 tok/s | Runs great | 43Fair quality (C) · 43 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 34.5 tok/s | Runs welltight fit |
| Qwen2.5 Coder 7B Instruct | 55Good quality (B) · 55 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 57.8 tok/s | Runs great | 46Fair quality (C) · 46 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 12.2 tok/s | Runs well |
| Phi-4 Multimodal Instruct | 44Fair quality (C) · 44 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 79.1 tok/s | Runs great | 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 13.0 tok/s | Runs well |
| Jamba Reasoning 3B | 44Fair quality (C) · 44 of 100: how capable the model is at F16, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 63.5 tok/s | Runs great | 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 22.5 tok/s | Runs well |
| LFM2.5-VL 3B | 44Fair quality (C) · 44 of 100: how capable the model is at F16, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 65.1 tok/s | Runs great | 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 23.1 tok/s | Runs well |
| Command R+ 104B (04-2024) | 53Fair quality (C) · 53 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 10.8 tok/s | Runs well | 44Fair quality (C) · 44 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 6/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| North Micro Vision | 44Fair quality (C) · 44 of 100: how capable the model is at F16, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 81.9 tok/s | Runs great | 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 29.1 tok/s | Runs well |
| LFM2.5 2.6B | 44Fair quality (C) · 44 of 100: how capable the model is at F16, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 75.2 tok/s | Runs great | 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 26.7 tok/s | Runs well |
| Aya 23 35B | 47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 24.7 tok/s | Runs well | 35Basic quality (D) · 35 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 4/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| DeepSeek V2 Lite Chat | 40Fair quality (C) · 40 of 100: how capable the model is at F16, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 61.8 tok/s | Runs great | 33Basic quality (D) · 33 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 34.5 tok/s | Runs welltight fit |
| MiMo V2.6 Pro (1T-A42B) | 68Good quality (B) · 68 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 26.0 tok/s | Slowfar over GPU memory limit | 68Good quality (B) · 68 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Olmo 3 7B Think | 54Fair quality (C) · 54 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 60.4 tok/s | Runs great | 45Fair quality (C) · 45 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 12.8 tok/s | Runs well |
| OpenReasoning Nemotron 7B | 54Fair quality (C) · 54 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 57.8 tok/s | Runs great | 45Fair quality (C) · 45 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 12.2 tok/s | Runs well |
| Llama 3.2 11B Vision Instruct | 42Fair quality (C) · 42 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 41.3 tok/s | Runs great | 29Basic quality (D) · 29 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 11.1 tok/s | Runs well |
| Yi 1.5 34B Chat | 47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 25.2 tok/s | Runs well | 35Basic quality (D) · 35 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 4/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Mistral-NeMo-Minitron 8B Instruct | 42Fair quality (C) · 42 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 52.4 tok/s | Runs great | 35Basic quality (D) · 35 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 11.1 tok/s | Runs well |
| Hunyuan 4B Instruct | 42Fair quality (C) · 42 of 100: how capable the model is at F16, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 48.1 tok/s | Runs great | 39Basic quality (D) · 39 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 17.1 tok/s | Runs well |
| Llama 3 8B Instruct | 42Fair quality (C) · 42 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 54.9 tok/s | Runs great | 35Basic quality (D) · 35 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 11.6 tok/s | Runs well |
| Kimi K2 Instruct 0905 (1T) | 63Good quality (B) · 63 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 33.2 tok/s | Slowfar over GPU memory limit | 63Good quality (B) · 63 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Command R7B Arabic | 42Fair quality (C) · 42 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 54.9 tok/s | Runs great | 35Basic quality (D) · 35 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 11.6 tok/s | Runs well |
| SmolLM3 3B | 42Fair quality (C) · 42 of 100: how capable the model is at F16, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 65.9 tok/s | Runs great | 39Basic quality (D) · 39 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 23.4 tok/s | Runs well |
| MiMo V2.5 Pro (1T-A42B) | 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 26.0 tok/s | Slowfar over GPU memory limit | 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Jamba2 3B | 42Fair quality (C) · 42 of 100: how capable the model is at F16, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 67.0 tok/s | Runs great | 39Basic quality (D) · 39 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 23.8 tok/s | Runs well |
| Foundation-Sec 8B Reasoning | 52Fair quality (C) · 52 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 54.9 tok/s | Runs great | 43Fair quality (C) · 43 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 11.6 tok/s | Runs well |
| Kimi K2 Instruct (July 2025) | 62Good quality (B) · 62 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 33.2 tok/s | Slowfar over GPU memory limit | 62Good quality (B) · 62 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| InternLM2 Chat 20B | 39Basic quality (D) · 39 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 43.6 tok/s | Runs great | 29Basic quality (D) · 29 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | 5.1 tok/s | Slowover GPU memory limit |
| Nous Hermes 2 SOLAR 10.7B | 40Fair quality (C) · 40 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 41.1 tok/s | Runs great | 28Basic quality (D) · 28 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 11.0 tok/s | Runs well |
| SOLAR 10.7B Instruct | 40Fair quality (C) · 40 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 41.1 tok/s | Runs great | 28Basic quality (D) · 28 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 11.0 tok/s | Runs well |
| Phi-3.5 Mini Instruct | 40Fair quality (C) · 40 of 100: how capable the model is at F16, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 53.2 tok/s | Runs great | 38Basic quality (D) · 38 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 18.9 tok/s | Runs well |
| AFM 4.5B | 40Fair quality (C) · 40 of 100: how capable the model is at F16, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 44.0 tok/s | Runs great | 38Basic quality (D) · 38 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 15.6 tok/s | Runs well |
| Apertus 8B Instruct | 40Fair quality (C) · 40 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 54.7 tok/s | Runs great | 33Basic quality (D) · 33 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 11.6 tok/s | Runs well |
| Aya Expanse 8B | 40Fair quality (C) · 40 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 54.9 tok/s | Runs great | 33Basic quality (D) · 33 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 11.6 tok/s | Runs well |
| Granite 3.0 8B Instruct | 40Fair quality (C) · 40 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 53.9 tok/s | Runs great | 33Basic quality (D) · 33 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 11.4 tok/s | Runs well |
| InternLM2.5 7B Chat | 40Fair quality (C) · 40 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 56.9 tok/s | Runs great | 33Basic quality (D) · 33 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 12.0 tok/s | Runs well |
| Mistral Large 4 (1T-A52B) | 67Good quality (B) · 67 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 21.4 tok/s | Slowfar over GPU memory limit | 67Good quality (B) · 67 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Granite 4.0 Micro (3B) | 40Fair quality (C) · 40 of 100: how capable the model is at F16, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 59.7 tok/s | Runs great | 38Basic quality (D) · 38 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 21.2 tok/s | Runs well |
| Qwen3.5 2B | 40Fair quality (C) · 40 of 100: how capable the model is at F16, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 89.5 tok/s | Runs great | 38Basic quality (D) · 38 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 31.8 tok/s | Runs great |
| Devstral 2 (123B) | 60Good quality (B) · 60 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 11.4 tok/s | Runs well | 60Good quality (B) · 60 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| TMax 2B | 40Fair quality (C) · 40 of 100: how capable the model is at F16, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 108 tok/s | Runs great | 38Basic quality (D) · 38 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 38.4 tok/s | Runs great |
| MiniCPM-V 4.7 1B | 40Fair quality (C) · 40 of 100: how capable the model is at F16, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 131 tok/s | Runs great | 38Basic quality (D) · 38 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 38.5 tok/s | Runs great |
| Qianfan-OCR 4B | 50Fair quality (C) · 50 of 100: how capable the model is at F16, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 42.9 tok/s | Runs great | 47Fair quality (C) · 47 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 15.2 tok/s | Runs well |
| LFM2 8B-A1B | 38Basic quality (D) · 38 of 100: how capable the model is at F16, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 85.0 tok/s | Runs great | 36Basic quality (D) · 36 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 41.4 tok/s | Runs great |
| DeepSeek R1 Distill Llama 8B | 50Fair quality (C) · 50 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 54.9 tok/s | Runs great | 42Fair quality (C) · 42 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 11.6 tok/s | Runs well |
| Fara 7B | 50Fair quality (C) · 50 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 53.2 tok/s | Runs great | 42Fair quality (C) · 42 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 11.3 tok/s | Runs well |
| Phi-3.5 Vision Instruct | 38Basic quality (D) · 38 of 100: how capable the model is at F16, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 48.9 tok/s | Runs great | 36Basic quality (D) · 36 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 17.4 tok/s | Runs well |
| Dolphin 2.9 Llama 3 8B | 38Basic quality (D) · 38 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 54.9 tok/s | Runs great | 31Basic quality (D) · 31 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 11.6 tok/s | Runs well |
| EXAONE 3.0 7.8B Instruct | 38Basic quality (D) · 38 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 56.4 tok/s | Runs great | 31Basic quality (D) · 31 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 11.9 tok/s | Runs well |
| Mistral 7B Instruct v0.3 | 38Basic quality (D) · 38 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 60.8 tok/s | Runs great | 31Basic quality (D) · 31 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 12.9 tok/s | Runs well |
| Zamba2 7B Instruct | 38Basic quality (D) · 38 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 59.9 tok/s | Runs great | 31Basic quality (D) · 31 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 12.7 tok/s | Runs well |
| MiniCPM3 4B | 38Basic quality (D) · 38 of 100: how capable the model is at F16, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 49.9 tok/s | Runs great | 36Basic quality (D) · 36 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 17.7 tok/s | Runs well |
| Falcon3 Mamba 7B Instruct | 38Basic quality (D) · 38 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 60.6 tok/s | Runs great | 31Basic quality (D) · 31 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 12.8 tok/s | Runs well |
| Granite 4.0 H Micro | 38Basic quality (D) · 38 of 100: how capable the model is at F16, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 63.7 tok/s | Runs great | 36Basic quality (D) · 36 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 22.6 tok/s | Runs well |
| Qwen1.5 32B Chat | 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 26.6 tok/s | Runs well | 30Basic quality (D) · 30 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 4/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| S5 nano | 37Basic quality (D) · 37 of 100: how capable the model is at F16, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 43.6 tok/s | Runs great | 35Basic quality (D) · 35 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 15.5 tok/s | Runs well |
| CogVLM2 Llama 3 Chat 19B | 36Basic quality (D) · 36 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 44.4 tok/s | Runs great | 26Basic quality (D) · 26 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 3/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| LightOnOCR-3 4B | 48Fair quality (C) · 48 of 100: how capable the model is at F16, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 44.7 tok/s | Runs great | 45Fair quality (C) · 45 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 15.9 tok/s | Runs well |
| LLaVA 1.6 34B | 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 24.9 tok/s | Runs well | 30Basic quality (D) · 30 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 4/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Chandra OCR 2 | 48Fair quality (C) · 48 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 83.2 tok/s | Runs great | 45Fair quality (C) · 45 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 13.6 tok/s | Runs well |
| Gemma 3n E2B | 36Basic quality (D) · 36 of 100: how capable the model is at F16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 102 tok/s | Runs great | 34Basic quality (D) · 34 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 36.1 tok/s | Runs great |
| Phi-3 Mini 128K Instruct | 36Basic quality (D) · 36 of 100: how capable the model is at F16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 53.2 tok/s | Runs great | 34Basic quality (D) · 34 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 18.9 tok/s | Runs well |
| Voxtral Mini 3B | 36Basic quality (D) · 36 of 100: how capable the model is at F16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 43.4 tok/s | Runs great | 34Basic quality (D) · 34 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 15.4 tok/s | Runs well |
| OLMo 2 7B Instruct | 36Basic quality (D) · 36 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 60.4 tok/s | Runs great | 30Basic quality (D) · 30 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 12.8 tok/s | Runs well |
| Yi 1.5 9B Chat | 36Basic quality (D) · 36 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 49.9 tok/s | Runs great | 30Basic quality (D) · 30 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 10.6 tok/s | Runs well |
| Phi-3 Mini 4K Instruct | 36Basic quality (D) · 36 of 100: how capable the model is at F16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 53.2 tok/s | Runs great | 34Basic quality (D) · 34 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 18.9 tok/s | Runs well |
| Hermes 2 Pro Mistral 7B | 36Basic quality (D) · 36 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 60.9 tok/s | Runs great | 30Basic quality (D) · 30 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 12.9 tok/s | Runs well |
| Starling LM 7B Beta | 36Basic quality (D) · 36 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 60.9 tok/s | Runs great | 30Basic quality (D) · 30 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 12.9 tok/s | Runs well |
| Ling 1T | 63Good quality (B) · 63 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 22.2 tok/s | Slowfar over GPU memory limit | 63Good quality (B) · 63 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Falcon-H1 3B Instruct | 36Basic quality (D) · 36 of 100: how capable the model is at F16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 64.5 tok/s | Runs great | 34Basic quality (D) · 34 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 22.9 tok/s | Runs well |
| Qwen3-VL 2B Instruct | 36Basic quality (D) · 36 of 100: how capable the model is at F16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 95.4 tok/s | Runs great | 34Basic quality (D) · 34 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 33.9 tok/s | Runs great |
| MiniCPM5 1B | 36Basic quality (D) · 36 of 100: how capable the model is at F16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 188 tok/s | Runs great | 35Basic quality (D) · 35 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 46.3 tok/s | Runs great |
| CodeGeeX4 All 9B | 46Fair quality (C) · 46 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 46.9 tok/s | Runs great | 32Basic quality (D) · 32 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 12.6 tok/s | Runs well |
| Ling 2.6 1T | 64Good quality (B) · 64 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 17.9 tok/s | Slowfar over GPU memory limit | 64Good quality (B) · 64 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| OpenChat 3.5 (0106) | 35Basic quality (D) · 35 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 60.9 tok/s | Runs great | 29Basic quality (D) · 29 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 12.9 tok/s | Runs well |
| Qwen1.5 14B Chat | 33Basic quality (D) · 33 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 42.4 tok/s | Runs great | 23Basic quality (D) · 23 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 3/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Phi-4 Mini Reasoning (3.8B) | 46Fair quality (C) · 46 of 100: how capable the model is at F16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 52.9 tok/s | Runs great | 43Fair quality (C) · 43 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 18.8 tok/s | Runs well |
| DeepSeek R1 Distill Qwen 7B | 46Fair quality (C) · 46 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 57.8 tok/s | Runs great | 38Basic quality (D) · 38 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 12.2 tok/s | Runs well |
| Llama 3.2 3B Instruct | 35Basic quality (D) · 35 of 100: how capable the model is at F16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 63.3 tok/s | Runs great | 33Basic quality (D) · 33 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 22.5 tok/s | Runs well |
| Raon 1 nano | 35Basic quality (D) · 35 of 100: how capable the model is at F16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 57.5 tok/s | Runs great | 33Basic quality (D) · 33 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 20.4 tok/s | Runs well |
| Phi-3 Vision 128K Instruct | 34Basic quality (D) · 34 of 100: how capable the model is at F16, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 48.9 tok/s | Runs great | 32Basic quality (D) · 32 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 17.4 tok/s | Runs well |
| RecurrentGemma 9B IT | 34Basic quality (D) · 34 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 45.8 tok/s | Runs great | 23Basic quality (D) · 23 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 12.3 tok/s | Runs well |
| WeVisOne-VL 4B | 45Fair quality (C) · 45 of 100: how capable the model is at F16, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 42.1 tok/s | Runs great | 42Fair quality (C) · 42 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 14.9 tok/s | Runs well |
| Aya 23 8B | 34Basic quality (D) · 34 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 54.9 tok/s | Runs great | 28Basic quality (D) · 28 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 11.6 tok/s | Runs well |
| Mistral 7B Instruct v0.2 | 34Basic quality (D) · 34 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 60.9 tok/s | Runs great | 28Basic quality (D) · 28 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 12.9 tok/s | Runs well |
| Nous Hermes 2 Mistral 7B DPO | 34Basic quality (D) · 34 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 60.9 tok/s | Runs great | 28Basic quality (D) · 28 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 12.9 tok/s | Runs well |
| OpenChat 3.5 (1210) | 34Basic quality (D) · 34 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 60.9 tok/s | Runs great | 28Basic quality (D) · 28 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 12.9 tok/s | Runs well |
| Qwen2.5-VL 3B Instruct | 34Basic quality (D) · 34 of 100: how capable the model is at F16, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 54.2 tok/s | Runs great | 32Basic quality (D) · 32 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 19.2 tok/s | Runs well |
| Qwen2.5-Math 72B Instruct | 55Good quality (B) · 55 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 11.9 tok/s | Runs well | 40Fair quality (C) · 40 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 5/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen2.5 3B Instruct | 34Basic quality (D) · 34 of 100: how capable the model is at F16, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 65.7 tok/s | Runs great | 32Basic quality (D) · 32 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 23.3 tok/s | Runs well |
| LFM2 2.6B | 34Basic quality (D) · 34 of 100: how capable the model is at F16, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 79.0 tok/s | Runs great | 28Basic quality (D) · 28 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 36.3 tok/s | Runs great |
| Qwen1.5 110B Chat | 43Fair quality (C) · 43 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 10.1 tok/s | Runs well | 36Basic quality (D) · 36 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 4/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Kimi K2.7 Code (1T) | 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 33.2 tok/s | Slowfar over GPU memory limit | 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Yi 34B Chat | 38Basic quality (D) · 38 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 25.2 tok/s | Runs well | 28Basic quality (D) · 28 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 3/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Unlimited OCR 2 (3B-A0.6B) | 44Fair quality (C) · 44 of 100: how capable the model is at F16, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 136 tok/s | Runs great | 44Fair quality (C) · 44 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 49.6 tok/s | Runs great |
| Kimi K2 Thinking (1T) | 65Good quality (B) · 65 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | 33.2 tok/s | Slowfar over GPU memory limit | 65Good quality (B) · 65 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| OpenHermes 2.5 Mistral 7B | 32Basic quality (D) · 32 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | 60.9 tok/s | Runs great | 26Basic quality (D) · 26 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 12.9 tok/s | Runs well |
| Starling LM 7B Alpha | 32Basic quality (D) · 32 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | 60.9 tok/s | Runs great | 26Basic quality (D) · 26 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 12.9 tok/s | Runs well |
| Hermes 3 Llama 3.2 3B | 32Basic quality (D) · 32 of 100: how capable the model is at F16, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | 63.3 tok/s | Runs great | 30Basic quality (D) · 30 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 22.5 tok/s | Runs well |
| LFM2.5 1.2B Instruct | 32Basic quality (D) · 32 of 100: how capable the model is at F16, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | 174 tok/s | Runs great | 31Basic quality (D) · 31 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 42.8 tok/s | Runs great |
| MiniCPM-OCR 1.0 1.3B | 42Fair quality (C) · 42 of 100: how capable the model is at F16, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 156 tok/s | Runs great | 41Fair quality (C) · 41 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 38.5 tok/s | Runs great |
| Mistral 7B Instruct v0.1 | 30Basic quality (D) · 30 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 60.9 tok/s | Runs great | 25Basic quality (D) · 25 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 12.9 tok/s | Runs well |
| PaddleOCR-VL 1.6 0.9B | 42Fair quality (C) · 42 of 100: how capable the model is at F16, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 212 tok/s | Runs great | 41Fair quality (C) · 41 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 52.1 tok/s | Runs great |
| Falcon Mamba 7B Instruct | 30Basic quality (D) · 30 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 60.6 tok/s | Runs great | 25Basic quality (D) · 25 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 12.8 tok/s | Runs well |
| Falcon 3 3B Instruct | 30Basic quality (D) · 30 of 100: how capable the model is at F16, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 62.9 tok/s | Runs great | 25Basic quality (D) · 25 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 28.9 tok/s | Runs well |
| Qwen2.5-Omni 3B | 30Basic quality (D) · 30 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 79.6 tok/s | Runs great | 28Basic quality (D) · 28 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 13.0 tok/s | Runs well |
| Granite 3.2 2B Instruct | 30Basic quality (D) · 30 of 100: how capable the model is at F16, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 80.3 tok/s | Runs great | 25Basic quality (D) · 25 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 36.9 tok/s | Runs great |
| Granite 3.3 2B | 30Basic quality (D) · 30 of 100: how capable the model is at F16, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 80.3 tok/s | Runs great | 25Basic quality (D) · 25 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 36.9 tok/s | Runs great |
| EXAONE 3.5 2.4B | 30Basic quality (D) · 30 of 100: how capable the model is at F16, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 84.3 tok/s | Runs great | 25Basic quality (D) · 25 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 38.7 tok/s | Runs great |
| Qwen3 1.7B | 30Basic quality (D) · 30 of 100: how capable the model is at F16, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 100 tok/s | Runs great | 28Basic quality (D) · 28 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 35.5 tok/s | Runs great |
| DeepSeek MoE 16B Chat | 26Basic quality (D) · 26 of 100: how capable the model is at F16, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 55.1 tok/s | Runs great | 18Basic quality (D) · 18 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 37.4 tok/s | Runs great |
| EXAONE 4.0 1.2B | 30Basic quality (D) · 30 of 100: how capable the model is at F16, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 159 tok/s | Runs great | 29Basic quality (D) · 29 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 39.1 tok/s | Runs great |
| DeepSeek Prover V2 7B | 40Fair quality (C) · 40 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 63.8 tok/s | Runs great | 33Basic quality (D) · 33 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 13.5 tok/s | Runs well |
| Yi Coder 9B Chat | 40Fair quality (C) · 40 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 49.9 tok/s | Runs great | 33Basic quality (D) · 33 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 10.6 tok/s | Runs well |
| Unlimited OCR (3B-A0.6B) | 40Fair quality (C) · 40 of 100: how capable the model is at F16, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 136 tok/s | Runs great | 40Fair quality (C) · 40 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 49.6 tok/s | Runs great |
| Gemma 1.1 7B IT | 28Basic quality (D) · 28 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 51.6 tok/s | Runs great | 19Basic quality (D) · 19 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 13.9 tok/s | Runs well |
| Orca 2 13B | 27Basic quality (D) · 27 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 46.1 tok/s | Runs great | 19Basic quality (D) · 19 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 9.1 tok/s | Slowtight fit |
| TeleOCR | 40Fair quality (C) · 40 of 100: how capable the model is at F16, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 143 tok/s | Runs great | 38Basic quality (D) · 38 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 42.0 tok/s | Runs great |
| StarCoder2 15B Instruct | 38Basic quality (D) · 38 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 44.8 tok/s | Runs great | 33Basic quality (D) · 33 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 5.8 tok/s | Slowtight fit |
| DeepSeek LLM 67B Chat | 38Basic quality (D) · 38 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 12.8 tok/s | Runs well | 28Basic quality (D) · 28 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 3/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| LLaVA 1.6 Mistral 7B | 28Basic quality (D) · 28 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 58.2 tok/s | Runs great | 23Basic quality (D) · 23 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 12.3 tok/s | Runs well |
| Neural Chat 7B v3.3 | 28Basic quality (D) · 28 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 60.9 tok/s | Runs great | 23Basic quality (D) · 23 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 12.9 tok/s | Runs well |
| StableLM 2 12B Chat | 27Basic quality (D) · 27 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 49.5 tok/s | Runs great | 23Basic quality (D) · 23 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 7.7 tok/s | Slow |
| Yi 1.5 6B Chat | 28Basic quality (D) · 28 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 72.7 tok/s | Runs great | 26Basic quality (D) · 26 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | 11.9 tok/s | Runs well |
| Granite 3.1 2B Instruct | 28Basic quality (D) · 28 of 100: how capable the model is at F16, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 80.3 tok/s | Runs great | 23Basic quality (D) · 23 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 36.9 tok/s | Runs great |
| Gemma 2 2B | 28Basic quality (D) · 28 of 100: how capable the model is at F16, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 77.8 tok/s | Runs great | 23Basic quality (D) · 23 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | 35.7 tok/s | Runs great |
| InternLM2 Math Plus 20B | 38Basic quality (D) · 38 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 43.6 tok/s | Runs great | 28Basic quality (D) · 28 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 42/100 (quality, fit, speed and memory use) sets the ranking. | 5.1 tok/s | Slowover GPU memory limit |
| Granite 4.0 H 1B | 28Basic quality (D) · 28 of 100: how capable the model is at F16, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 139 tok/s | Runs great | 27Basic quality (D) · 27 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 40.8 tok/s | Runs great |
| LLM4Decompile 9B V2 | 38Basic quality (D) · 38 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 49.9 tok/s | Runs great | 31Basic quality (D) · 31 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | 10.6 tok/s | Runs well |
| Qwen1.5 MoE A2.7B Chat | 24Basic quality (D) · 24 of 100: how capable the model is at F16, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 56.6 tok/s | Runs great | 16Basic quality (D) · 16 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 38.6 tok/s | Runs great |
| OLMoE 1B-7B Instruct | 26Basic quality (D) · 26 of 100: how capable the model is at F16, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 92.8 tok/s | Runs great | 25Basic quality (D) · 25 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 39.5 tok/s | Runs great |
| WeVisOne VL 2B | 27Basic quality (D) · 27 of 100: how capable the model is at F16, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 83.2 tok/s | Runs great | 22Basic quality (D) · 22 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 38.2 tok/s | Runs great |
| Gemma 7B IT | 26Basic quality (D) · 26 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 51.6 tok/s | Runs great | 18Basic quality (D) · 18 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 13.9 tok/s | Runs well |
| Baichuan 2 13B Chat | 25Basic quality (D) · 25 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 43.2 tok/s | Runs great | 18Basic quality (D) · 18 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 8.5 tok/s | Slowtight fit |
| Qwen1.5 7B Chat | 26Basic quality (D) · 26 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 57.1 tok/s | Runs great | 18Basic quality (D) · 18 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 15.3 tok/s | Runs well |
| Llama 2 13B Chat | 25Basic quality (D) · 25 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 46.1 tok/s | Runs great | 18Basic quality (D) · 18 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 9.1 tok/s | Slowtight fit |
| DeepSeek Coder 33B Instruct | 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 26.0 tok/s | Runs well | 30Basic quality (D) · 30 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 4/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Zephyr 7B Beta | 26Basic quality (D) · 26 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 60.9 tok/s | Runs great | 21Basic quality (D) · 21 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | 12.9 tok/s | Runs well |
| Granite Vision 3.2 2B | 26Basic quality (D) · 26 of 100: how capable the model is at F16, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 68.2 tok/s | Runs great | 18Basic quality (D) · 18 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 39.8 tok/s | Runs great |
| Granite 3.0 2B Instruct | 26Basic quality (D) · 26 of 100: how capable the model is at F16, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 77.2 tok/s | Runs great | 21Basic quality (D) · 21 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 35.5 tok/s | Runs great |
| Falcon-H1 1.5B Instruct | 26Basic quality (D) · 26 of 100: how capable the model is at F16, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 131 tok/s | Runs great | 24Basic quality (D) · 24 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | 46.5 tok/s | Runs great |
| LFM2 1.2B | 26Basic quality (D) · 26 of 100: how capable the model is at F16, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 174 tok/s | Runs great | 25Basic quality (D) · 25 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 42.8 tok/s | Runs great |
| Qwen3.5 0.8B | 26Basic quality (D) · 26 of 100: how capable the model is at F16, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 233 tok/s | Runs great | 26Basic quality (D) · 26 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 42.2 tok/s | Runs great |
| Qwen2.5-Math 7B Instruct | 36Basic quality (D) · 36 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 57.8 tok/s | Runs great | 30Basic quality (D) · 30 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 12.2 tok/s | Runs well |
| Codestral Mamba 7B | 36Basic quality (D) · 36 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 60.5 tok/s | Runs great | 30Basic quality (D) · 30 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 12.8 tok/s | Runs well |
| Qwen2.5 Coder 3B Instruct | 36Basic quality (D) · 36 of 100: how capable the model is at F16, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 65.7 tok/s | Runs great | 34Basic quality (D) · 34 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 23.3 tok/s | Runs well |
| Llama 2 70B Chat | 34Basic quality (D) · 34 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 12.6 tok/s | Runs well | 25Basic quality (D) · 25 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 3/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Falcon 180B Chat | 32Basic quality (D) · 32 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 4.8 tok/s | Slow | 23Basic quality (D) · 23 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 3/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| SmolVLM2 2.2B Instruct | 24Basic quality (D) · 24 of 100: how capable the model is at F16, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 90.3 tok/s | Runs great | 20Basic quality (D) · 20 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 41.5 tok/s | Runs great |
| Granite Code 20B Instruct | 34Basic quality (D) · 34 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 43.1 tok/s | Runs great | 25Basic quality (D) · 25 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 5.9 tok/s | Slowtight fit |
| Qwen2-VL 2B Instruct | 24Basic quality (D) · 24 of 100: how capable the model is at F16, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 91.9 tok/s | Runs great | 20Basic quality (D) · 20 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 42.2 tok/s | Runs great |
| CodeGemma 7B | 34Basic quality (D) · 34 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 51.6 tok/s | Runs great | 28Basic quality (D) · 28 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 10.9 tok/s | Runs welltight fit |
| Qwen2.5 1.5B Instruct | 24Basic quality (D) · 24 of 100: how capable the model is at F16, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 132 tok/s | Runs great | 22Basic quality (D) · 22 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 46.8 tok/s | Runs great |
| LFM2-VL 1.6B | 24Basic quality (D) · 24 of 100: how capable the model is at F16, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 129 tok/s | Runs great | 22Basic quality (D) · 22 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 45.7 tok/s | Runs great |
| DeepSeek Coder 6.7B Instruct | 34Basic quality (D) · 34 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 65.4 tok/s | Runs great | 28Basic quality (D) · 28 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | 13.8 tok/s | Runs well |
| Mathstral 7B | 34Basic quality (D) · 34 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 60.8 tok/s | Runs great | 28Basic quality (D) · 28 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 12.9 tok/s | Runs well |
| LLaVA 1.5 13B | 21Basic quality (D) · 21 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 45.0 tok/s | Runs great | 15Basic quality (D) · 15 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 8.9 tok/s | Slowtight fit |
| WizardLM 13B v1.2 | 21Basic quality (D) · 21 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 46.1 tok/s | Runs great | 15Basic quality (D) · 15 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 9.1 tok/s | Slowtight fit |
| Orca 2 7B | 22Basic quality (D) · 22 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 65.4 tok/s | Runs great | 20Basic quality (D) · 20 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 10.7 tok/s | Runs well |
| Nemotron Mini 4B Instruct | 22Basic quality (D) · 22 of 100: how capable the model is at F16, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 48.5 tok/s | Runs great | 15Basic quality (D) · 15 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 28.3 tok/s | Runs well |
| Granite Code 34B Instruct | 38Basic quality (D) · 38 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 25.7 tok/s | Runs well | 28Basic quality (D) · 28 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 3/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| SmolVLM 2.2B Instruct | 22Basic quality (D) · 22 of 100: how capable the model is at F16, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 90.3 tok/s | Runs great | 18Basic quality (D) · 18 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 41.5 tok/s | Runs great |
| Zamba2 2.7B Instruct | 22Basic quality (D) · 22 of 100: how capable the model is at F16, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 76.4 tok/s | Runs great | 18Basic quality (D) · 18 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 35.1 tok/s | Runs great |
| Hunyuan 1.8B Instruct | 22Basic quality (D) · 22 of 100: how capable the model is at F16, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 113 tok/s | Runs great | 20Basic quality (D) · 20 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | 40.3 tok/s | Runs great |
| LLM4Decompile 6.7B V2 | 32Basic quality (D) · 32 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 65.4 tok/s | Runs great | 26Basic quality (D) · 26 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 13.8 tok/s | Runs well |
| Llama 3.2 1B Instruct | 22Basic quality (D) · 22 of 100: how capable the model is at F16, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 164 tok/s | Runs great | 21Basic quality (D) · 21 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 40.4 tok/s | Runs great |
| Granite Code 8B Instruct (128K) | 32Basic quality (D) · 32 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 54.7 tok/s | Runs great | 26Basic quality (D) · 26 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 11.6 tok/s | Runs well |
| InternLM2 Math Plus 7B | 32Basic quality (D) · 32 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 56.9 tok/s | Runs great | 26Basic quality (D) · 26 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 12.0 tok/s | Runs well |
| Code Llama 70B Instruct | 41Fair quality (C) · 41 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 12.6 tok/s | Runs well | 30Basic quality (D) · 30 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 4/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Vicuna 13B v1.5 | 19Basic quality (D) · 19 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 46.1 tok/s | Runs great | 14Basic quality (D) · 14 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 9.1 tok/s | Slowtight fit |
| StarCoder2 15B (base) | 31Basic quality (D) · 31 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 44.8 tok/s | Runs great | 26Basic quality (D) · 26 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 5.8 tok/s | Slowtight fit |
| Baichuan 2 7B Chat | 20Basic quality (D) · 20 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 58.7 tok/s | Runs great | 16Basic quality (D) · 16 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 12.4 tok/s | Runs well |
| Code Llama 34B Instruct | 36Basic quality (D) · 36 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 25.7 tok/s | Runs well | 26Basic quality (D) · 26 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 3/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Llama 2 7B Chat | 20Basic quality (D) · 20 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 65.4 tok/s | Runs great | 16Basic quality (D) · 16 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | 13.8 tok/s | Runs well |
| Granite 3.1 3B-A800M Instruct | 20Basic quality (D) · 20 of 100: how capable the model is at F16, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 120 tok/s | Runs great | 19Basic quality (D) · 19 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 51.0 tok/s | Runs great |
| Code Llama 13B Instruct | 29Basic quality (D) · 29 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 46.1 tok/s | Runs great | 21Basic quality (D) · 21 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 35/100 (quality, fit, speed and memory use) sets the ranking. | 7.7 tok/s | Slowfar over GPU memory limit |
| Yi 6B Chat | 20Basic quality (D) · 20 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 72.7 tok/s | Runs great | 16Basic quality (D) · 16 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 15.4 tok/s | Runs well |
| SmolLM2 1.7B Instruct | 20Basic quality (D) · 20 of 100: how capable the model is at F16, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 119 tok/s | Runs great | 19Basic quality (D) · 19 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 42.2 tok/s | Runs great |
| SQLCoder 7B v2 | 30Basic quality (D) · 30 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 65.4 tok/s | Runs great | 25Basic quality (D) · 25 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 13.8 tok/s | Runs well |
| Gemma 3 1B | 20Basic quality (D) · 20 of 100: how capable the model is at F16, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 203 tok/s | Runs great | 19Basic quality (D) · 19 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 50.1 tok/s | Runs great |
| DeepSeek LLM 7B Chat | 18Basic quality (D) · 18 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | 63.8 tok/s | Runs great | 15Basic quality (D) · 15 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 13.5 tok/s | Runs well |
| Granite 3.0 3B-A800M Instruct | 18Basic quality (D) · 18 of 100: how capable the model is at F16, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | 120 tok/s | Runs great | 17Basic quality (D) · 17 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 51.0 tok/s | Runs great |
| InternLM2.5 1.8B Chat | 18Basic quality (D) · 18 of 100: how capable the model is at F16, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | 107 tok/s | Runs great | 17Basic quality (D) · 17 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 38.2 tok/s | Runs great |
| LFM2 700M | 18Basic quality (D) · 18 of 100: how capable the model is at F16, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | 274 tok/s | Runs great | 18Basic quality (D) · 18 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 49.6 tok/s | Runs great |
| GLiDE 2B | 17Basic quality (D) · 17 of 100: how capable the model is at F16, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | 108 tok/s | Runs great | 16Basic quality (D) · 16 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 38.4 tok/s | Runs great |
| LLaVA 1.5 7B | 16Basic quality (D) · 16 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | 62.4 tok/s | Runs great | 11Basic quality (D) · 11 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 16.8 tok/s | Runs well |
| StableLM Zephyr 3B | 16Basic quality (D) · 16 of 100: how capable the model is at F16, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | 72.5 tok/s | Runs great | 11Basic quality (D) · 11 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 42.3 tok/s | Runs great |
| Phi-2 2.7B | 16Basic quality (D) · 16 of 100: how capable the model is at F16, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | 73.1 tok/s | Runs great | 11Basic quality (D) · 11 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 42.6 tok/s | Runs great |
| Falcon 3 1B Instruct | 16Basic quality (D) · 16 of 100: how capable the model is at F16, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | 122 tok/s | Runs great | 15Basic quality (D) · 15 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 43.2 tok/s | Runs great |
| Spark X2.6 1.7B | 16Basic quality (D) · 16 of 100: how capable the model is at F16, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | 119 tok/s | Runs great | 15Basic quality (D) · 15 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 42.2 tok/s | Runs great |
| Qwen2 1.5B Instruct | 16Basic quality (D) · 16 of 100: how capable the model is at F16, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | 132 tok/s | Runs great | 15Basic quality (D) · 15 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 46.8 tok/s | Runs great |
| WizardMath 7B v1.1 | 26Basic quality (D) · 26 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | 60.9 tok/s | Runs great | 21Basic quality (D) · 21 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 12.9 tok/s | Runs well |
| StarCoder 15B | 25Basic quality (D) · 25 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | 45.2 tok/s | Runs great | 18Basic quality (D) · 18 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | 7.5 tok/s | Slow |
| DeepScaleR 1.5B Preview | 26Basic quality (D) · 26 of 100: how capable the model is at F16, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | 114 tok/s | Runs great | 24Basic quality (D) · 24 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 40.5 tok/s | Runs great |
| Qwen2.5 Coder 1.5B Instruct | 26Basic quality (D) · 26 of 100: how capable the model is at F16, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | 132 tok/s | Runs great | 24Basic quality (D) · 24 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 46.8 tok/s | Runs great |
| Vicuna 7B v1.5 | 14Basic quality (D) · 14 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | 65.4 tok/s | Runs great | 9Basic quality (D) · 9 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 17.6 tok/s | Runs well |
| OLMo 7B Instruct | 14Basic quality (D) · 14 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | 64.0 tok/s | Runs great | 9Basic quality (D) · 9 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | 17.2 tok/s | Runs well |
| SmolLM 1.7B Instruct | 14Basic quality (D) · 14 of 100: how capable the model is at F16, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | 119 tok/s | Runs great | 13Basic quality (D) · 13 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 42.2 tok/s | Runs great |
| OLMo 2 1B Instruct | 14Basic quality (D) · 14 of 100: how capable the model is at F16, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | 137 tok/s | Runs great | 13Basic quality (D) · 13 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 40.3 tok/s | Runs great |
| Code Llama 7B Instruct | 24Basic quality (D) · 24 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | 65.4 tok/s | Runs great | 20Basic quality (D) · 20 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 13.8 tok/s | Runs well |
| Qwen3 0.6B | 14Basic quality (D) · 14 of 100: how capable the model is at F16, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | 271 tok/s | Runs great | 14Basic quality (D) · 14 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 49.0 tok/s | Runs great |
| Falcon-H1 0.5B Instruct | 14Basic quality (D) · 14 of 100: how capable the model is at F16, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | 391 tok/s | Runs great | 14Basic quality (D) · 14 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | 70.6 tok/s | Runs great |
| LFM2.5 350M | 14Basic quality (D) · 14 of 100: how capable the model is at F16, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | 580 tok/s | Runs great | 14Basic quality (D) · 14 of 100: how capable the model is at F16, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 48.4 tok/s | Runs great |
| StarCoder2 7B | 24Basic quality (D) · 24 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | 61.5 tok/s | Runs great | 22Basic quality (D) · 22 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 10.1 tok/s | Runs well |
| InternLM 7B | 12Basic quality (D) · 12 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | 60.2 tok/s | Runs great | 8Basic quality (D) · 8 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | 16.2 tok/s | Runs well |
| Gemma 2B IT | 12Basic quality (D) · 12 of 100: how capable the model is at F16, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | 80.9 tok/s | Runs great | 10Basic quality (D) · 10 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | 37.2 tok/s | Runs great |
| LFM2-VL 450M | 12Basic quality (D) · 12 of 100: how capable the model is at F16, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | 451 tok/s | Runs great | 12Basic quality (D) · 12 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | 81.6 tok/s | Runs great |
| LFM2 350M | 12Basic quality (D) · 12 of 100: how capable the model is at F16, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | 580 tok/s | Runs great | 12Basic quality (D) · 12 of 100: how capable the model is at F16, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | 48.4 tok/s | Runs great |
| Granite 4.0 H 350M | 12Basic quality (D) · 12 of 100: how capable the model is at F16, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | 597 tok/s | Runs great | 12Basic quality (D) · 12 of 100: how capable the model is at F16, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | 49.8 tok/s | Runs great |
| DeepSeek R1 Distill Qwen 1.5B | 22Basic quality (D) · 22 of 100: how capable the model is at F16, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | 114 tok/s | Runs great | 20Basic quality (D) · 20 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | 40.5 tok/s | Runs great |
| GLiDE 0.8B | 11Basic quality (D) · 11 of 100: how capable the model is at F16, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | 271 tok/s | Runs great | 11Basic quality (D) · 11 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | 49.0 tok/s | Runs great |
| Falcon 40B Instruct | 11Basic quality (D) · 11 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | 34.0 tok/s | Runs great | 11Basic quality (D) · 11 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 1/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| StableLM 2 1.6B Chat | 10Basic quality (D) · 10 of 100: how capable the model is at F16, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | 124 tok/s | Runs great | 9Basic quality (D) · 9 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 44.0 tok/s | Runs great |
| StableLM 2 Zephyr 1.6B | 10Basic quality (D) · 10 of 100: how capable the model is at F16, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | 124 tok/s | Runs great | 9Basic quality (D) · 9 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 44.0 tok/s | Runs great |
| d1 2B | 21Basic quality (D) · 21 of 100: how capable the model is at F16, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | 106 tok/s | Runs great | 19Basic quality (D) · 19 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | 37.6 tok/s | Runs great |
| SmolVLM 500M Instruct | 10Basic quality (D) · 10 of 100: how capable the model is at F16, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | 398 tok/s | Runs great | 10Basic quality (D) · 10 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | 72.0 tok/s | Runs great |
| Granite Code 3B Instruct (128K) | 20Basic quality (D) · 20 of 100: how capable the model is at F16, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | 58.4 tok/s | Runs great | 14Basic quality (D) · 14 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 34.0 tok/s | Runs great |
| Qwen2.5 0.5B Instruct | 10Basic quality (D) · 10 of 100: how capable the model is at F16, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | 415 tok/s | Runs great | 10Basic quality (D) · 10 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | 75.0 tok/s | Runs great |
| LFM2.5 230M | 10Basic quality (D) · 10 of 100: how capable the model is at F16, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | 883 tok/s | Runs great | 10Basic quality (D) · 10 of 100: how capable the model is at F16, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | 73.6 tok/s | Runs great |
| Stable Code Instruct 3B | 20Basic quality (D) · 20 of 100: how capable the model is at F16, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | 72.5 tok/s | Runs great | 14Basic quality (D) · 14 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 42.3 tok/s | Runs great |
| Qwen2.5-Math 1.5B Instruct | 20Basic quality (D) · 20 of 100: how capable the model is at F16, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | 132 tok/s | Runs great | 19Basic quality (D) · 19 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | 46.8 tok/s | Runs great |
| Hunyuan 0.5B Instruct | 8Basic quality (D) · 8 of 100: how capable the model is at F16, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | 376 tok/s | Runs great | 8Basic quality (D) · 8 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | 68.0 tok/s | Runs great |
| SmolLM2 360M Instruct | 8Basic quality (D) · 8 of 100: how capable the model is at F16, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | 564 tok/s | Runs great | 8Basic quality (D) · 8 of 100: how capable the model is at F16, whatever it runs on. Match score 54/100 (quality, fit, speed and memory use) sets the ranking. | 47.0 tok/s | Runs great |
| StarCoder2 3B | 18Basic quality (D) · 18 of 100: how capable the model is at F16, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | 67.0 tok/s | Runs great | 12Basic quality (D) · 12 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | 39.1 tok/s | Runs great |
| Falcon 7B Instruct | 6Basic quality (D) · 6 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | 61.0 tok/s | Runs great | 4Basic quality (D) · 4 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | 16.4 tok/s | Runs well |
| TinyLlama 1.1B Chat | 6Basic quality (D) · 6 of 100: how capable the model is at F16, whatever it runs on. Match score 49/100 (quality, fit, speed and memory use) sets the ranking. | 185 tok/s | Runs great | 5Basic quality (D) · 5 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | 45.5 tok/s | Runs great |
| Qwen2 0.5B Instruct | 6Basic quality (D) · 6 of 100: how capable the model is at F16, whatever it runs on. Match score 49/100 (quality, fit, speed and memory use) sets the ranking. | 415 tok/s | Runs great | 6Basic quality (D) · 6 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | 75.0 tok/s | Runs great |
| ERNIE 4.5 0.3B | 6Basic quality (D) · 6 of 100: how capable the model is at F16, whatever it runs on. Match score 49/100 (quality, fit, speed and memory use) sets the ranking. | 564 tok/s | Runs great | 6Basic quality (D) · 6 of 100: how capable the model is at F16, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | 47.0 tok/s | Runs great |
| SmolVLM 256M Instruct | 6Basic quality (D) · 6 of 100: how capable the model is at F16, whatever it runs on. Match score 49/100 (quality, fit, speed and memory use) sets the ranking. | 781 tok/s | Runs great | 6Basic quality (D) · 6 of 100: how capable the model is at F16, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | 65.1 tok/s | Runs great |
| Gemma 3 270M | 6Basic quality (D) · 6 of 100: how capable the model is at F16, whatever it runs on. Match score 49/100 (quality, fit, speed and memory use) sets the ranking. | 752 tok/s | Runs great | 6Basic quality (D) · 6 of 100: how capable the model is at F16, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | 62.7 tok/s | Runs great |
| Yi Coder 1.5B Chat | 16Basic quality (D) · 16 of 100: how capable the model is at F16, whatever it runs on. Match score 49/100 (quality, fit, speed and memory use) sets the ranking. | 137 tok/s | Runs great | 15Basic quality (D) · 15 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | 40.3 tok/s | Runs great |
| SmolLM 360M Instruct | 5Basic quality (D) · 5 of 100: how capable the model is at F16, whatever it runs on. Match score 49/100 (quality, fit, speed and memory use) sets the ranking. | 564 tok/s | Runs great | 5Basic quality (D) · 5 of 100: how capable the model is at F16, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | 47.0 tok/s | Runs great |
| surya ocr 2.5 | 15Basic quality (D) · 15 of 100: how capable the model is at F16, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | 294 tok/s | Runs great | 15Basic quality (D) · 15 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 51/100 (quality, fit, speed and memory use) sets the ranking. | 53.2 tok/s | Runs great |
| SmolLM2 135M Instruct | 4Basic quality (D) · 4 of 100: how capable the model is at F16, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | 1,562 tok/s | Runs great | 4Basic quality (D) · 4 of 100: how capable the model is at F16, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | 130 tok/s | Runs great |
| CodeGemma 2B | 14Basic quality (D) · 14 of 100: how capable the model is at F16, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | 80.9 tok/s | Runs great | 11Basic quality (D) · 11 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | 37.2 tok/s | Runs great |
| Falcon-H1 Tiny 90M Instruct | 3Basic quality (D) · 3 of 100: how capable the model is at F16, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | 2,257 tok/s | Runs great | 3Basic quality (D) · 3 of 100: how capable the model is at F16, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | 188 tok/s | Runs great |
| SmolLM 135M Instruct | 2Basic quality (D) · 2 of 100: how capable the model is at F16, whatever it runs on. Match score 47/100 (quality, fit, speed and memory use) sets the ranking. | 1,562 tok/s | Runs great | 2Basic quality (D) · 2 of 100: how capable the model is at F16, whatever it runs on. Match score 49/100 (quality, fit, speed and memory use) sets the ranking. | 130 tok/s | Runs great |
| DeepSeek Coder 1.3B Instruct | 12Basic quality (D) · 12 of 100: how capable the model is at F16, whatever it runs on. Match score 47/100 (quality, fit, speed and memory use) sets the ranking. | 150 tok/s | Runs great | 11Basic quality (D) · 11 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | 44.1 tok/s | Runs great |
| LLM4Decompile 1.3B V2 | 12Basic quality (D) · 12 of 100: how capable the model is at F16, whatever it runs on. Match score 47/100 (quality, fit, speed and memory use) sets the ranking. | 150 tok/s | Runs great | 11Basic quality (D) · 11 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | 44.1 tok/s | Runs great |
| Qwen2.5 Coder 0.5B Instruct | 8Basic quality (D) · 8 of 100: how capable the model is at F16, whatever it runs on. Match score 45/100 (quality, fit, speed and memory use) sets the ranking. | 415 tok/s | Runs great | 8Basic quality (D) · 8 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 48/100 (quality, fit, speed and memory use) sets the ranking. | 75.0 tok/s | Runs great |
| Kimi K3.1 Pro (2.8T) | 69Good quality (B) · 69 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run | 69Good quality (B) · 69 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen3.8 Max (2.4T-A95B) | 68Good quality (B) · 68 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run | 68Good quality (B) · 68 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Kimi K3 (2.8T) | 68Good quality (B) · 68 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run | 68Good quality (B) · 68 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| LongCat 2.5 (1.6T-A48B) | 68Good quality (B) · 68 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run | 68Good quality (B) · 68 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| DeepSeek V4 Pro (1.6T-A49B) | 67Good quality (B) · 67 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run | 67Good quality (B) · 67 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| DeepSeek V4 Pro (April 2026) | 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run | 66Good quality (B) · 66 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
Image models
61 of 61 image models run on the Apple M3 Ultra (512 GB). 384 GB usable for models.
53 of 61 image models run on the Apple M1 (16 GB). 10.7 GB usable for models.
| Model | QualityM3 Ultra | SpeedM3 Ultra | RunsM3 Ultra | QualityM1 | SpeedM1 | RunsM1 |
|---|---|---|---|---|---|---|
| FLUX.2 [klein] 9B | 84Great quality (A) · 84 of 100: how capable the model is at FP16, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | 7.6 s | Runs great | 78Great quality (A) · 78 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | 58 s | Slowencoder unloaded after use |
| Krea 2 Turbo | 92Excellent quality (A+) · 92 of 100: how capable the model is at FP16, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | 18 s | Runs well | 90Excellent quality (A+) · 90 of 100: how capable the model is at FP8, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 2.1 min | SlowCPU offload |
| Z-Image Turbo | 84Great quality (A) · 84 of 100: how capable the model is at FP16, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | 11 s | Runs great | 78Great quality (A) · 78 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | 76 s | Slowtight fit |
| HunyuanImage 3.0 (80B-A13B) | 90Excellent quality (A+) · 90 of 100: how capable the model is at FP16, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | 4.2 min | Very slow | 88Excellent quality (A+) · 88 of 100: how capable the model is at FP8, whatever it runs on. Match score 11/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Cosmos3-Super Text2Image (64B) | 92Excellent quality (A+) · 92 of 100: how capable the model is at FP16, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | 2.8 min | Very slow | 90Excellent quality (A+) · 90 of 100: how capable the model is at FP8, whatever it runs on. Match score 12/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| MiniMax H3 Image | 91Excellent quality (A+) · 91 of 100: how capable the model is at FP8, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | 38 s | Runs well | 91Excellent quality (A+) · 91 of 100: how capable the model is at FP8, whatever it runs on. Match score 12/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| ERNIE-Image Turbo | 85Excellent quality (A+) · 85 of 100: how capable the model is at FP16, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | 14 s | Runs well | 79Great quality (A) · 79 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 1.7 min | Slowencoder unloaded after use |
| FLUX.2 [dev] | 92Excellent quality (A+) · 92 of 100: how capable the model is at FP16, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | 2.5 min | Very slow | 85Excellent quality (A+) · 85 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 23.8 min | Very slowtight fit · CPU offload |
| FLUX 3 Image Klein | 91Excellent quality (A+) · 91 of 100: how capable the model is at FP16, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | 24 s | Runs well | 84Great quality (A) · 84 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | 2.9 min | Very slowtight fit |
| FLUX.2 [klein] 4B | 78Great quality (A) · 78 of 100: how capable the model is at FP16, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | 4.2 s | Runs great | 76Great quality (A) · 76 of 100: how capable the model is at FP8, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | 23 s | Runs welltight fit |
| Qwen-Image 2.1 Turbo | 89Excellent quality (A+) · 89 of 100: how capable the model is at FP16, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 49 s | Slow | 82Great quality (A) · 82 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 11/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Ideogram 4 | 92Excellent quality (A+) · 92 of 100: how capable the model is at FP16, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 1.8 min | Slow | 90Excellent quality (A+) · 90 of 100: how capable the model is at FP8, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 12.4 min | Very slowCPU offload |
| FLUX.1 [schnell] | 70Great quality (A) · 70 of 100: how capable the model is at FP16, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 8.5 s | Runs great | 65Good quality (B) · 65 of 100: how capable the model is at NF4, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 59 s | Slowtight fit · encoder unloaded after use |
| Fibo 2.0 Dev | 93Excellent quality (A+) · 93 of 100: how capable the model is at FP16, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 1.6 min | Slow | 93Excellent quality (A+) · 93 of 100: how capable the model is at FP16, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 12.4 min | Very slowCPU offload |
| Qwen-Image 2.1 | 93Excellent quality (A+) · 93 of 100: how capable the model is at FP16, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 4.2 min | Very slow | 86Excellent quality (A+) · 86 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 11/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Ming-Image 0.1 Design | 86Excellent quality (A+) · 86 of 100: how capable the model is at FP16, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 69 s | Slow | 84Great quality (A) · 84 of 100: how capable the model is at FP8, whatever it runs on. Match score 11/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen-Image 2512 | 90Excellent quality (A+) · 90 of 100: how capable the model is at FP16, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 6.3 min | Very slow | 83Great quality (A) · 83 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 11/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| ERNIE-Image | 88Excellent quality (A+) · 88 of 100: how capable the model is at FP16, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 1.6 min | Slow | 81Great quality (A) · 81 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 11.7 min | Very slowencoder unloaded after use |
| Qwen-Image | 88Excellent quality (A+) · 88 of 100: how capable the model is at FP16, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 6.3 min | Very slow | 86Excellent quality (A+) · 86 of 100: how capable the model is at FP8, whatever it runs on. Match score 11/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| FLUX.1 Krea [dev] | 82Great quality (A) · 82 of 100: how capable the model is at FP16, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 51 s | Slow | 76Great quality (A) · 76 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 6.4 min | Very slowtight fit · encoder unloaded after use |
| HiDream-I1 Dev | 80Great quality (A) · 80 of 100: how capable the model is at FP16, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 72 s | Slow | 78Great quality (A) · 78 of 100: how capable the model is at FP8, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 8.4 min | Very slowCPU offload |
| FireRed-Image-Edit 1.1 | 87Excellent quality (A+) · 87 of 100: how capable the model is at FP16, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 5.6 min | Very slow | 80Great quality (A) · 80 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 52.9 min | Very slowCPU offload |
| HiDream-I1 Full | 84Great quality (A) · 84 of 100: how capable the model is at FP16, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 3.2 min | Very slow | 82Great quality (A) · 82 of 100: how capable the model is at FP8, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 22.3 min | Very slowCPU offload |
| Qwen-Image-Edit 2511 | 86Excellent quality (A+) · 86 of 100: how capable the model is at FP16, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 5.6 min | Very slow | 79Great quality (A) · 79 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 52.9 min | Very slowCPU offload |
| FLUX.1 [dev] | 80Great quality (A) · 80 of 100: how capable the model is at FP16, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 51 s | Slow | 74Great quality (A) · 74 of 100: how capable the model is at NF4, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 5.9 min | Very slowtight fit · encoder unloaded after use |
| MOSS-Image SFT | 82Great quality (A) · 82 of 100: how capable the model is at FP16, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 59 s | Slow | 81Great quality (A) · 81 of 100: how capable the model is at Q8 GGUF, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 7.2 min | Very slowencoder unloaded after use |
| Stable Diffusion 3.5 Large Turbo | 62Good quality (B) · 62 of 100: how capable the model is at FP16, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 6.8 s | Runs great | 57Good quality (B) · 57 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 51 s | Slowencoder unloaded after use |
| FLUX.1 Kontext [dev] | 82Great quality (A) · 82 of 100: how capable the model is at FP16, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 1.6 min | Slow | 81Great quality (A) · 81 of 100: how capable the model is at Q8 GGUF, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 13.9 min | Very slowCPU offload |
| Z-Image | 82Great quality (A) · 82 of 100: how capable the model is at FP16, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 72 s | Slow | 76Great quality (A) · 76 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 8.6 min | Very slowtight fit |
| HiDream-I1 Fast | 74Great quality (A) · 74 of 100: how capable the model is at FP8, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 38 s | Runs well | 74Great quality (A) · 74 of 100: how capable the model is at FP8, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 5.0 min | Very slowCPU offload |
| FLUX.1 Fill [dev] | 76Great quality (A) · 76 of 100: how capable the model is at FP16, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 53 s | Slow | 74Great quality (A) · 74 of 100: how capable the model is at FP8, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 6.2 min | Very slowCPU offload |
| Chroma1-HD | 78Great quality (A) · 78 of 100: how capable the model is at FP16, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 85 s | Slow | 72Great quality (A) · 72 of 100: how capable the model is at NF4, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 9.8 min | Very slowencoder unloaded after use |
| DreamShaper XL v2 Turbo | 58Good quality (B) · 58 of 100: how capable the model is at FP16, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 5.5 s | Runs great | 58Good quality (B) · 58 of 100: how capable the model is at FP16, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 33 s | Runs well |
| SDXL-Lightning (4-step) | 54Fair quality (C) · 54 of 100: how capable the model is at FP16, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 3.0 s | Runs great | 52Fair quality (C) · 52 of 100: how capable the model is at FP8, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 16 s | Runs well |
| Kolors | 58Good quality (B) · 58 of 100: how capable the model is at FP8, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 15 s | Runs well | 60Good quality (B) · 60 of 100: how capable the model is at FP16, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 1.8 min | Slowencoder unloaded after use |
| Iris-3B | 72Great quality (A) · 72 of 100: how capable the model is at FP16, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 68 s | Slow | 72Great quality (A) · 72 of 100: how capable the model is at FP16, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 7.1 min | Very slowencoder unloaded after use |
| Juggernaut XL v9 | 60Good quality (B) · 60 of 100: how capable the model is at FP8, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 17 s | Runs well | 62Good quality (B) · 62 of 100: how capable the model is at FP16, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 1.9 min | Slow |
| RealVisXL V5.0 | 60Good quality (B) · 60 of 100: how capable the model is at FP8, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 17 s | Runs well | 62Good quality (B) · 62 of 100: how capable the model is at FP16, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 1.9 min | Slow |
| CogView4 6B | 70Great quality (A) · 70 of 100: how capable the model is at FP16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 76 s | Slow | 68Good quality (B) · 68 of 100: how capable the model is at FP8, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 7.2 min | Very slowencoder unloaded after use |
| Animagine XL 4.0 | 58Good quality (B) · 58 of 100: how capable the model is at FP8, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 16 s | Runs well | 60Good quality (B) · 60 of 100: how capable the model is at FP16, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 1.8 min | Slow |
| Illustrious XL v0.1 | 58Good quality (B) · 58 of 100: how capable the model is at FP8, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 16 s | Runs well | 60Good quality (B) · 60 of 100: how capable the model is at FP16, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 1.8 min | Slow |
| Stable Diffusion 3.5 Large | 66Good quality (B) · 66 of 100: how capable the model is at FP16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 47 s | Slow | 61Good quality (B) · 61 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 5.9 min | Very slowencoder unloaded after use |
| Pony Diffusion V6 XL | 56Good quality (B) · 56 of 100: how capable the model is at FP8, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 14 s | Runs well | 58Good quality (B) · 58 of 100: how capable the model is at FP16, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 1.6 min | Slow |
| PixArt-Σ XL 1024 | 50Fair quality (C) · 50 of 100: how capable the model is at FP16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 6.3 s | Runs great | 50Fair quality (C) · 50 of 100: how capable the model is at FP16, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 40 s | Slowencoder unloaded after use |
| Lumina-Image 2.0 | 62Good quality (B) · 62 of 100: how capable the model is at FP16, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 30 s | Runs well | 60Good quality (B) · 60 of 100: how capable the model is at FP8, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 2.7 min | Very slow |
| Stable Diffusion 3 Medium | 50Fair quality (C) · 50 of 100: how capable the model is at FP8, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 12 s | Runs great | 50Fair quality (C) · 50 of 100: how capable the model is at FP8, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 71 s | Slowtight fit |
| Animagine XL 3.1 | 54Fair quality (C) · 54 of 100: how capable the model is at FP8, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | 16 s | Runs well | 56Good quality (B) · 56 of 100: how capable the model is at FP16, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 1.8 min | Slow |
| Stable Diffusion 3.5 Medium | 58Good quality (B) · 58 of 100: how capable the model is at FP16, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | 25 s | Runs well | 56Good quality (B) · 56 of 100: how capable the model is at FP8, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 2.3 min | Very slowtight fit |
| OmniGen2 | 66Good quality (B) · 66 of 100: how capable the model is at FP16, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | 68 s | Slow | 64Good quality (B) · 64 of 100: how capable the model is at FP8, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 6.1 min | Very slowtight fit |
| Stable Diffusion XL 1.0 | 53Fair quality (C) · 53 of 100: how capable the model is at FP8, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | 17 s | Runs well | 55Good quality (B) · 55 of 100: how capable the model is at FP16, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 1.9 min | Slow |
| Playground v2.5 | 60Good quality (B) · 60 of 100: how capable the model is at FP16, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 32 s | Runs well | 60Good quality (B) · 60 of 100: how capable the model is at FP16, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 3.2 min | Very slow |
| Stable Cascade | 50Fair quality (C) · 50 of 100: how capable the model is at FP8, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 19 s | Runs well | 50Fair quality (C) · 50 of 100: how capable the model is at FP8, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 1.9 min | Slow |
| Realistic Vision V5.1 | 42Fair quality (C) · 42 of 100: how capable the model is at FP16, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 5.1 s | Runs great | 42Fair quality (C) · 42 of 100: how capable the model is at FP16, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 31 s | Runs well |
| SDXL Turbo | 40Fair quality (C) · 40 of 100: how capable the model is at FP16, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 0.6 s | Runs great | 40Fair quality (C) · 40 of 100: how capable the model is at FP16, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 3.8 s | Runs great |
| AuraFlow v0.3 | 58Good quality (B) · 58 of 100: how capable the model is at FP16, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 59 s | Slow | 53Fair quality (C) · 53 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 7.1 min | Very slow |
| Hunyuan-DiT v1.2 | 52Fair quality (C) · 52 of 100: how capable the model is at FP16, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 25 s | Runs well | 52Fair quality (C) · 52 of 100: how capable the model is at FP16, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 2.5 min | Very slowtight fit |
| DreamShaper 8 | 40Fair quality (C) · 40 of 100: how capable the model is at FP16, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 5.1 s | Runs great | 40Fair quality (C) · 40 of 100: how capable the model is at FP16, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 31 s | Runs well |
| Stable Diffusion 1.5 | 35Basic quality (D) · 35 of 100: how capable the model is at FP16, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | 5.1 s | Runs great | 35Basic quality (D) · 35 of 100: how capable the model is at FP16, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | 31 s | Runs well |
| Kandinsky 3 | 48Fair quality (C) · 48 of 100: how capable the model is at FP16, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | 38 s | Runs well | 48Fair quality (C) · 48 of 100: how capable the model is at FP16, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 4.0 min | Very slowencoder unloaded after use |
| LCM DreamShaper v7 | 30Basic quality (D) · 30 of 100: how capable the model is at FP16, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | 1.1 s | Runs great | 30Basic quality (D) · 30 of 100: how capable the model is at FP16, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 6.4 s | Runs great |
| SD Turbo | 28Basic quality (D) · 28 of 100: how capable the model is at FP16, whatever it runs on. Match score 52/100 (quality, fit, speed and memory use) sets the ranking. | 0.3 s | Runs great | 28Basic quality (D) · 28 of 100: how capable the model is at FP16, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 2.0 s | Runs great |
Video models
43 of 43 video models run on the Apple M3 Ultra (512 GB). 384 GB usable for models.
12 of 43 video models run on the Apple M1 (16 GB). 10.7 GB usable for models.
| Model | QualityM3 Ultra | SpeedM3 Ultra | RunsM3 Ultra | QualityM1 | SpeedM1 | RunsM1 |
|---|---|---|---|---|---|---|
| LTX-2.5 (22B, video + audio) | 88Excellent quality (A+) · 88 of 100: how capable the model is at FP16, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | 6.4 min | Runs well | 81Great quality (A) · 81 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 11/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| MiniMax H3 (33B, video + audio) | 90Excellent quality (A+) · 90 of 100: how capable the model is at FP16, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 73.3 min | Very slow | 83Great quality (A) · 83 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 11/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Cosmos3-Super Image2Video (64B) | 88Excellent quality (A+) · 88 of 100: how capable the model is at FP16, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 2.6 h | Very slow | 88Excellent quality (A+) · 88 of 100: how capable the model is at FP16, whatever it runs on. Match score 11/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| LTX-2.3 | 86Excellent quality (A+) · 86 of 100: how capable the model is at FP16, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 28.2 min | Slow | 79Great quality (A) · 79 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 10/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Kandinsky 6.0 Pro (29B, video + audio) | 88Excellent quality (A+) · 88 of 100: how capable the model is at FP16, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 3.5 h | Very slow | 88Excellent quality (A+) · 88 of 100: how capable the model is at FP16, whatever it runs on. Match score 11/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| LTX-2 (19B, video + audio) | 82Great quality (A) · 82 of 100: how capable the model is at FP16, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 28.2 min | Slow | 76Great quality (A) · 76 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 10/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| LTX-Video 13B 0.9.8 (distilled) | 66Good quality (B) · 66 of 100: how capable the model is at FP8, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 3.8 min | Runs great | 66Good quality (B) · 66 of 100: how capable the model is at FP8, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Bernini-R A14B (video generation + editing) | 85Excellent quality (A+) · 85 of 100: how capable the model is at FP16, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 50.8 min | Very slow | 83Great quality (A) · 83 of 100: how capable the model is at FP8, whatever it runs on. Match score 11/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Wan 2.2 I2V A14B | 86Excellent quality (A+) · 86 of 100: how capable the model is at FP16, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 1.9 h | Very slow | 79Great quality (A) · 79 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 10/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Wan 2.2 T2V A14B | 86Excellent quality (A+) · 86 of 100: how capable the model is at FP16, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 1.9 h | Very slow | 79Great quality (A) · 79 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 10/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| HunyuanVideo 1.5 (8.3B) | 84Great quality (A) · 84 of 100: how capable the model is at FP16, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 2.1 h | Very slow | 82Great quality (A) · 82 of 100: how capable the model is at FP8, whatever it runs on. Match score 11/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| SCAIL-2 14B (character animation) | 80Great quality (A) · 80 of 100: how capable the model is at FP16, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 84.6 min | Very slow | 78Great quality (A) · 78 of 100: how capable the model is at FP8, whatever it runs on. Match score 10/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Wan 2.2 Animate 14B | 78Great quality (A) · 78 of 100: how capable the model is at FP16, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 2.2 h | Very slow | 72Great quality (A) · 72 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| FramePack I2V (HunyuanVideo 13B) | 74Great quality (A) · 74 of 100: how capable the model is at FP16, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 28.2 min | Slow | 72Great quality (A) · 72 of 100: how capable the model is at FP8, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Wan 2.1 VACE 14B | 78Great quality (A) · 78 of 100: how capable the model is at FP16, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 2.6 h | Very slow | 72Great quality (A) · 72 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| HY-OmniWeaving (8.3B, unified video) | 76Great quality (A) · 76 of 100: how capable the model is at FP16, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 37.4 min | Slow | 74Great quality (A) · 74 of 100: how capable the model is at FP8, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 4.4 h | Very slowCPU offload |
| Wan 2.1 I2V 14B 720P | 78Great quality (A) · 78 of 100: how capable the model is at FP16, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 2.4 h | Very slow | 72Great quality (A) · 72 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Wan 2.1 T2V 14B | 78Great quality (A) · 78 of 100: how capable the model is at FP16, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 2.4 h | Very slow | 72Great quality (A) · 72 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| SkyReels V2 T2V 14B 720P | 78Great quality (A) · 78 of 100: how capable the model is at FP16, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 2.8 h | Very slow | 76Great quality (A) · 76 of 100: how capable the model is at FP8, whatever it runs on. Match score 10/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Step-Video T2V (30B) | 76Great quality (A) · 76 of 100: how capable the model is at FP16, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 3.6 h | Very slow | 74Great quality (A) · 74 of 100: how capable the model is at FP8, whatever it runs on. Match score 10/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Wan 2.2 S2V 14B (speech to video) | 76Great quality (A) · 76 of 100: how capable the model is at FP16, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 70.5 min | Very slow | 70Great quality (A) · 70 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 11.0 h | Very slowCPU offload |
| Wan-Dancer 14B (music to dance) | 76Great quality (A) · 76 of 100: how capable the model is at FP16, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 5.2 h | Very slow | 74Great quality (A) · 74 of 100: how capable the model is at FP8, whatever it runs on. Match score 10/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Wan 2.1 I2V 14B 480P | 74Great quality (A) · 74 of 100: how capable the model is at FP16, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 63.5 min | Very slow | 68Good quality (B) · 68 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 9.9 h | Very slowCPU offload |
| SkyReels V2 T2V 14B 540P | 74Great quality (A) · 74 of 100: how capable the model is at FP16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 1.6 h | Very slow | 72Great quality (A) · 72 of 100: how capable the model is at FP8, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| LTX-Video 2B | 50Fair quality (C) · 50 of 100: how capable the model is at FP16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 1.8 min | Runs great | 50Fair quality (C) · 50 of 100: how capable the model is at FP16, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 11.1 min | Runs wellencoder unloaded after use |
| Wan 2.2 TI2V 5B | 70Great quality (A) · 70 of 100: how capable the model is at FP16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 36.7 min | Slow | 65Good quality (B) · 65 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| HunyuanVideo 13B | 72Great quality (A) · 72 of 100: how capable the model is at FP16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 2.8 h | Very slow | 66Good quality (B) · 66 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| HunyuanVideo I2V 13B | 72Great quality (A) · 72 of 100: how capable the model is at FP16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 2.8 h | Very slow | 66Good quality (B) · 66 of 100: how capable the model is at Q4 GGUF, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Kandinsky 6.0 Video Lite (3B, video + audio) | 68Good quality (B) · 68 of 100: how capable the model is at FP16, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 21.9 min | Slow | 68Good quality (B) · 68 of 100: how capable the model is at FP16, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 2.8 h | Very slowCPU offload |
| Cosmos-Predict2 14B Video2World | 70Great quality (A) · 70 of 100: how capable the model is at FP16, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 1.9 h | Very slow | 68Good quality (B) · 68 of 100: how capable the model is at FP8, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Open-Sora 2.0 (11B) | 70Great quality (A) · 70 of 100: how capable the model is at FP16, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 3.5 h | Very slow | 70Great quality (A) · 70 of 100: how capable the model is at FP16, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Cosmos-Predict2 2B Video2World | 58Good quality (B) · 58 of 100: how capable the model is at FP16, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 17.6 min | Slow | 56Good quality (B) · 56 of 100: how capable the model is at FP8, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Cosmos 1.0 Diffusion 7B Text2World | 60Good quality (B) · 60 of 100: how capable the model is at FP16, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 38.8 min | Slow | 60Good quality (B) · 60 of 100: how capable the model is at FP16, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| CogVideoX 2B | 44Fair quality (C) · 44 of 100: how capable the model is at FP8, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 3.8 min | Runs great | 45Fair quality (C) · 45 of 100: how capable the model is at FP16, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 26.7 min | Slowencoder unloaded after use |
| Mochi 1 Preview (10B) | 62Good quality (B) · 62 of 100: how capable the model is at FP16, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 77.6 min | Very slow | 60Good quality (B) · 60 of 100: how capable the model is at FP8, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| CogVideoX 1.5 5B | 60Good quality (B) · 60 of 100: how capable the model is at FP16, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. | 63.5 min | Very slow | 58Good quality (B) · 58 of 100: how capable the model is at FP8, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| CogVideoX 5B I2V | 53Fair quality (C) · 53 of 100: how capable the model is at FP8, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 11.4 min | Runs well | 53Fair quality (C) · 53 of 100: how capable the model is at FP8, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 72.1 min | Very slowtight fit · encoder unloaded after use |
| CogVideoX 5B | 53Fair quality (C) · 53 of 100: how capable the model is at FP8, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 11.4 min | Runs well | 53Fair quality (C) · 53 of 100: how capable the model is at FP8, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. | 72.1 min | Very slowtight fit · encoder unloaded after use |
| Wan 2.1 T2V 1.3B | 55Good quality (B) · 55 of 100: how capable the model is at FP16, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 16.9 min | Slow | 55Good quality (B) · 55 of 100: how capable the model is at FP16, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. | 1.8 h | Very slowencoder unloaded after use |
| Wan 2.1 VACE 1.3B | 55Good quality (B) · 55 of 100: how capable the model is at FP16, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. | 18.3 min | Slow | 53Fair quality (C) · 53 of 100: how capable the model is at FP8, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking. | 1.7 h | Very slowencoder unloaded after use |
| AnimateDiff v3 (SD 1.5) | 35Basic quality (D) · 35 of 100: how capable the model is at FP16, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking. | 85 s | Runs great | 35Basic quality (D) · 35 of 100: how capable the model is at FP16, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. | 8.5 min | Runs well |
| SkyReels V2 Diffusion Forcing 1.3B | 52Fair quality (C) · 52 of 100: how capable the model is at FP16, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking. | 21.2 min | Slow | 52Fair quality (C) · 52 of 100: how capable the model is at FP16, whatever it runs on. Match score 50/100 (quality, fit, speed and memory use) sets the ranking. | 2.8 h | Very slowtight fit · CPU offload |
| Allegro (2.8B) | 52Fair quality (C) · 52 of 100: how capable the model is at FP16, whatever it runs on. Match score 53/100 (quality, fit, speed and memory use) sets the ranking. | 84.6 min | Very slow | 52Fair quality (C) · 52 of 100: how capable the model is at FP16, whatever it runs on. Match score 7/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
Decision models
70 of 70 decision models run on the Apple M3 Ultra (512 GB). 384 GB usable for models.
64 of 70 decision models run on the Apple M1 (16 GB). 10.7 GB usable for models.
| Model | QualityM3 Ultra | SpeedM3 Ultra | RunsM3 Ultra | QualityM1 | SpeedM1 | RunsM1 |
|---|---|---|---|---|---|---|
| Jina Reranker v3.5 | 86Excellent quality (A+) · 86 of 100: how capable the model is at F16, whatever it runs on. Match score 93/100 (quality, fit, speed and memory use) sets the ranking. | 24 ms | Runs great | 86Excellent quality (A+) · 86 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. | 211 ms | Runs well |
| Qwen3 Reranker 4B | 90Excellent quality (A+) · 90 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 93/100 (quality, fit, speed and memory use) sets the ranking. | 130 ms | Runs great | 90Excellent quality (A+) · 90 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 84/100 (quality, fit, speed and memory use) sets the ranking. | 1.4 s | Very slow |
| Intern-Decision 4B | 84Great quality (A) · 84 of 100: how capable the model is at F16, whatever it runs on. Match score 92/100 (quality, fit, speed and memory use) sets the ranking. | 92 ms | Runs great | 84Great quality (A) · 84 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | 849 ms | Slow |
| Jina Reranker v3 | 84Great quality (A) · 84 of 100: how capable the model is at F16, whatever it runs on. Match score 92/100 (quality, fit, speed and memory use) sets the ranking. | 24 ms | Runs great | 84Great quality (A) · 84 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 88/100 (quality, fit, speed and memory use) sets the ranking. | 211 ms | Runs well |
| zerank-2 | 88Excellent quality (A+) · 88 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 92/100 (quality, fit, speed and memory use) sets the ranking. | 130 ms | Runs great | 88Excellent quality (A+) · 88 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | 1.4 s | Very slow |
| Shieldstral 1.0 3B | 86Excellent quality (A+) · 86 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 92/100 (quality, fit, speed and memory use) sets the ranking. | 125 ms | Runs great | 86Excellent quality (A+) · 86 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | 1.3 s | Slow |
| d1 3B | 82Great quality (A) · 82 of 100: how capable the model is at F16, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. | 65 ms | Runs great | 82Great quality (A) · 82 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | 584 ms | Slow |
| mxbai-rerank-large v2 | 82Great quality (A) · 82 of 100: how capable the model is at F16, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. | 56 ms | Runs great | 82Great quality (A) · 82 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | 535 ms | Slow |
| FRIDA-Decisions | 80Great quality (A) · 80 of 100: how capable the model is at F16, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. | 19 ms | Runs great | 80Great quality (A) · 80 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 88/100 (quality, fit, speed and memory use) sets the ranking. | 146 ms | Runs well |
| GLiNER2.5 Decide | 80Great quality (A) · 80 of 100: how capable the model is at F16, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. | 13 ms | Runs great | 80Great quality (A) · 80 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. | 89 ms | Runs great |
| LemonDecider 4B | 80Great quality (A) · 80 of 100: how capable the model is at F16, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. | 87 ms | Runs great | 80Great quality (A) · 80 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | 788 ms | Slow |
| Qwen3 Reranker 0.6B | 80Great quality (A) · 80 of 100: how capable the model is at F16, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. | 24 ms | Runs great | 80Great quality (A) · 80 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 86/100 (quality, fit, speed and memory use) sets the ranking. | 211 ms | Runs well |
| zerank-1 small | 80Great quality (A) · 80 of 100: how capable the model is at F16, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. | 62 ms | Runs great | 80Great quality (A) · 80 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | 597 ms | Slow |
| BGE Reranker v2 Gemma | 78Great quality (A) · 78 of 100: how capable the model is at F16, whatever it runs on. Match score 90/100 (quality, fit, speed and memory use) sets the ranking. | 89 ms | Runs great | 78Great quality (A) · 78 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | 870 ms | Slow |
| Qwen3 Reranker 8B | 90Excellent quality (A+) · 90 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. | 261 ms | Runs well | 89Excellent quality (A+) · 89 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | 2.8 s | Very slow |
| Llama Prompt Guard 2 86M | 76Great quality (A) · 76 of 100: how capable the model is at F16, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. | 14 ms | Runs great | 76Great quality (A) · 76 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. | 101 ms | Runs great |
| mxbai-rerank-base v2 | 76Great quality (A) · 76 of 100: how capable the model is at F16, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. | 21 ms | Runs great | 76Great quality (A) · 76 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 85/100 (quality, fit, speed and memory use) sets the ranking. | 173 ms | Runs well |
| GLiClass ModernBERT Large v3 | 74Great quality (A) · 74 of 100: how capable the model is at F16, whatever it runs on. Match score 88/100 (quality, fit, speed and memory use) sets the ranking. | 11 ms | Runs great | 74Great quality (A) · 74 of 100: how capable the model is at F16, whatever it runs on. Match score 88/100 (quality, fit, speed and memory use) sets the ranking. | 83 ms | Runs great |
| OpenAI Privacy Filter | 78Great quality (A) · 78 of 100: how capable the model is at F16, whatever it runs on. Match score 88/100 (quality, fit, speed and memory use) sets the ranking. | 4.9 ms | Runs great | 78Great quality (A) · 78 of 100: how capable the model is at F16, whatever it runs on. Match score 88/100 (quality, fit, speed and memory use) sets the ranking. | 14 ms | Runs great |
| BGE Reranker v2 M3 | 72Great quality (A) · 72 of 100: how capable the model is at F16, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. | 25 ms | Runs great | 72Great quality (A) · 72 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | 201 ms | Runs well |
| DeBERTa v3 Large Zero-shot v2.0 | 72Great quality (A) · 72 of 100: how capable the model is at F16, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. | 12 ms | Runs great | 72Great quality (A) · 72 of 100: how capable the model is at F16, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. | 91 ms | Runs great |
| DuoGuard 0.5B | 72Great quality (A) · 72 of 100: how capable the model is at F16, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. | 22 ms | Runs great | 72Great quality (A) · 72 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | 173 ms | Runs well |
| Skywork Reward V2 Qwen3 0.6B | 72Great quality (A) · 72 of 100: how capable the model is at F16, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. | 48 ms | Runs great | 72Great quality (A) · 72 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. | 418 ms | Slow |
| Skywork Reward V2 Llama 3.1 8B | 90Excellent quality (A+) · 90 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. | 478 ms | Slow | 90Excellent quality (A+) · 90 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | 5.2 s | Very slowtight fit |
| GTE Reranker ModernBERT Base | 70Great quality (A) · 70 of 100: how capable the model is at F16, whatever it runs on. Match score 86/100 (quality, fit, speed and memory use) sets the ranking. | 9.5 ms | Runs great | 70Great quality (A) · 70 of 100: how capable the model is at F16, whatever it runs on. Match score 86/100 (quality, fit, speed and memory use) sets the ranking. | 64 ms | Runs great |
| Jina Reranker v2 Base Multilingual | 70Great quality (A) · 70 of 100: how capable the model is at F16, whatever it runs on. Match score 86/100 (quality, fit, speed and memory use) sets the ranking. | 14 ms | Runs great | 70Great quality (A) · 70 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 86/100 (quality, fit, speed and memory use) sets the ranking. | 101 ms | Runs great |
| ModernBERT Large Zero-shot v2.0 | 70Great quality (A) · 70 of 100: how capable the model is at F16, whatever it runs on. Match score 86/100 (quality, fit, speed and memory use) sets the ranking. | 11 ms | Runs great | 70Great quality (A) · 70 of 100: how capable the model is at F16, whatever it runs on. Match score 86/100 (quality, fit, speed and memory use) sets the ranking. | 83 ms | Runs great |
| GLiDE 27B | 86Excellent quality (A+) · 86 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 86/100 (quality, fit, speed and memory use) sets the ranking. | 416 ms | Slow | 86Excellent quality (A+) · 86 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 11/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Qwen3Guard Gen 4B | 83Great quality (A) · 83 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 85/100 (quality, fit, speed and memory use) sets the ranking. | 176 ms | Runs well | 84Great quality (A) · 84 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | 2.1 s | Very slow |
| GLiGuard 300M | 68Good quality (B) · 68 of 100: how capable the model is at F16, whatever it runs on. Match score 85/100 (quality, fit, speed and memory use) sets the ranking. | 12 ms | Runs great | 68Good quality (B) · 68 of 100: how capable the model is at F16, whatever it runs on. Match score 85/100 (quality, fit, speed and memory use) sets the ranking. | 87 ms | Runs great |
| Granite Embedding Reranker English R2 | 68Good quality (B) · 68 of 100: how capable the model is at F16, whatever it runs on. Match score 85/100 (quality, fit, speed and memory use) sets the ranking. | 9.5 ms | Runs great | 68Good quality (B) · 68 of 100: how capable the model is at F16, whatever it runs on. Match score 85/100 (quality, fit, speed and memory use) sets the ranking. | 64 ms | Runs great |
| Laya | 68Good quality (B) · 68 of 100: how capable the model is at F16, whatever it runs on. Match score 85/100 (quality, fit, speed and memory use) sets the ranking. | 12 ms | Runs great | 68Good quality (B) · 68 of 100: how capable the model is at F16, whatever it runs on. Match score 85/100 (quality, fit, speed and memory use) sets the ranking. | 87 ms | Runs great |
| Llama Prompt Guard 2 22M | 68Good quality (B) · 68 of 100: how capable the model is at F16, whatever it runs on. Match score 85/100 (quality, fit, speed and memory use) sets the ranking. | 6.6 ms | Runs great | 68Good quality (B) · 68 of 100: how capable the model is at F16, whatever it runs on. Match score 85/100 (quality, fit, speed and memory use) sets the ranking. | 32 ms | Runs great |
| Qwen3Guard Gen 0.6B | 74Great quality (A) · 74 of 100: how capable the model is at F16, whatever it runs on. Match score 85/100 (quality, fit, speed and memory use) sets the ranking. | 48 ms | Runs great | 73Great quality (A) · 73 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | 332 ms | Runs well |
| GLiClass Base v3 | 66Good quality (B) · 66 of 100: how capable the model is at F16, whatever it runs on. Match score 84/100 (quality, fit, speed and memory use) sets the ranking. | 7.5 ms | Runs great | 66Good quality (B) · 66 of 100: how capable the model is at F16, whatever it runs on. Match score 84/100 (quality, fit, speed and memory use) sets the ranking. | 42 ms | Runs great |
| ProtectAI DeBERTa v3 Prompt Injection v2 | 66Good quality (B) · 66 of 100: how capable the model is at F16, whatever it runs on. Match score 84/100 (quality, fit, speed and memory use) sets the ranking. | 11 ms | Runs great | 66Good quality (B) · 66 of 100: how capable the model is at F16, whatever it runs on. Match score 84/100 (quality, fit, speed and memory use) sets the ranking. | 75 ms | Runs great |
| Skywork Reward V2 Qwen3 8B | 84Great quality (A) · 84 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 84/100 (quality, fit, speed and memory use) sets the ranking. | 483 ms | Slow | 84Great quality (A) · 84 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | 5.2 s | Very slowtight fit |
| DeBERTa v3 Base Zero-shot v2.0 | 64Good quality (B) · 64 of 100: how capable the model is at F16, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | 7.3 ms | Runs great | 64Good quality (B) · 64 of 100: how capable the model is at F16, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | 40 ms | Runs great |
| Qwen3Guard Gen 8B | 86Excellent quality (A+) · 86 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | 354 ms | Runs well | 85Excellent quality (A+) · 85 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | 3.6 s | Very slow |
| GLiNER2 Base v1 | 68Good quality (B) · 68 of 100: how capable the model is at F16, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | 7.8 ms | Runs great | 68Good quality (B) · 68 of 100: how capable the model is at F16, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. | 46 ms | Runs great |
| NuExtract 2.0 2B | 66Good quality (B) · 66 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | 97 ms | Runs great | 67Good quality (B) · 67 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 1.2 s | Slow |
| HHEM 2.1 Open | 62Good quality (B) · 62 of 100: how capable the model is at F16, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | 12 ms | Runs great | 62Good quality (B) · 62 of 100: how capable the model is at F16, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | 91 ms | Runs great |
| Llama 3.1 Nemotron Safety Guard 8B v3 | 84Great quality (A) · 84 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | 347 ms | Runs well | 83Great quality (A) · 83 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | 3.5 s | Very slow |
| LettuceDetect Large | 66Good quality (B) · 66 of 100: how capable the model is at F16, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | 33 ms | Runs great | 66Good quality (B) · 66 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | 280 ms | Runs well |
| Granite Guardian 4.1 8B | 84Great quality (A) · 84 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | 362 ms | Runs well | 83Great quality (A) · 83 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. | 3.7 s | Very slow |
| NuExtract 2.0 4B | 72Great quality (A) · 72 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | 162 ms | Runs well | 75Great quality (A) · 75 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 2.4 s | Very slow |
| Laya Multilingual | 60Good quality (B) · 60 of 100: how capable the model is at F16, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | 9.8 ms | Runs great | 60Good quality (B) · 60 of 100: how capable the model is at F16, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. | 68 ms | Runs great |
| WildGuard 7B | 79Great quality (A) · 79 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | 287 ms | Runs well | 79Great quality (A) · 79 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 3.2 s | Very slow |
| ShieldGemma 2B | 58Good quality (B) · 58 of 100: how capable the model is at F16, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. | 93 ms | Runs great | 58Good quality (B) · 58 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 904 ms | Slow |
| Granite Guardian 3.3 8B | 80Great quality (A) · 80 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | 353 ms | Runs well | 79Great quality (A) · 79 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 3.6 s | Very slow |
| GLiNER Multi v2.1 | 62Good quality (B) · 62 of 100: how capable the model is at F16, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | 9.3 ms | Runs great | 62Good quality (B) · 62 of 100: how capable the model is at F16, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | 62 ms | Runs great |
| BART Large MNLI | 56Good quality (B) · 56 of 100: how capable the model is at F16, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | 11 ms | Runs great | 56Good quality (B) · 56 of 100: how capable the model is at F16, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. | 85 ms | Runs great |
| ms-marco MiniLM L6 v2 | 52Fair quality (C) · 52 of 100: how capable the model is at F16, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | 4.8 ms | Runs great | 52Fair quality (C) · 52 of 100: how capable the model is at F16, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. | 13 ms | Runs great |
| ShieldGemma 9B | 64Good quality (B) · 64 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | 294 ms | Runs well | 62Good quality (B) · 62 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 3.1 s | Very slow |
| Llama Guard 3 1B | 56Good quality (B) · 56 of 100: how capable the model is at F16, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking. | 92 ms | Runs great | 55Good quality (B) · 55 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 660 ms | Slow |
| Llama Guard 3 8B | 67Good quality (B) · 67 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 75/100 (quality, fit, speed and memory use) sets the ranking. | 318 ms | Runs well | 67Good quality (B) · 67 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 3.5 s | Very slow |
| ShieldGemma 27B | 66Good quality (B) · 66 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 857 ms | Slow | 62Good quality (B) · 62 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| gpt-oss-safeguard 120B | 80Great quality (A) · 80 of 100: how capable the model is at MXFP4, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 1.6 s | Very slow | 80Great quality (A) · 80 of 100: how capable the model is at MXFP4, whatever it runs on. Match score 10/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| CompassJudger-2 32B | 86Excellent quality (A+) · 86 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 74/100 (quality, fit, speed and memory use) sets the ranking. | 13.1 s | Very slow | 81Great quality (A) · 81 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 11/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Nemotron Content Safety Reasoning 4B | 77Great quality (A) · 77 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 1.3 s | Slow | 78Great quality (A) · 78 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 20.1 s | Very slow |
| gpt-oss-safeguard 20B | 76Great quality (A) · 76 of 100: how capable the model is at MXFP4, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 1.3 s | Slow | 76Great quality (A) · 76 of 100: how capable the model is at MXFP4, whatever it runs on. Match score 10/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| Llama Guard 4 12B | 65Good quality (B) · 65 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 473 ms | Slow | 64Good quality (B) · 64 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 5.0 s | Very slowtight fit |
| DynaGuard 8B | 78Great quality (A) · 78 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. | 2.4 s | Very slow | 78Great quality (A) · 78 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 28.8 s | Very slow |
| Selene 1 Mini 8B | 79Great quality (A) · 79 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 2.5 s | Very slow | 79Great quality (A) · 79 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 70/100 (quality, fit, speed and memory use) sets the ranking. | 29.2 s | Very slow |
| DynaGuard 1.7B | 69Good quality (B) · 69 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. | 604 ms | Slow | 70Great quality (A) · 70 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 68/100 (quality, fit, speed and memory use) sets the ranking. | 9.5 s | Very slow |
| CompassJudger-2 7B | 77Great quality (A) · 77 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 71/100 (quality, fit, speed and memory use) sets the ranking. | 2.4 s | Very slow | 77Great quality (A) · 77 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 27.7 s | Very slow |
| Prometheus 2 8x7B | 74Great quality (A) · 74 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking. | 5.7 s | Very slow | 70Great quality (A) · 70 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. | — | Won't run |
| GLIDER | 65Good quality (B) · 65 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 1.2 s | Slow | 66Good quality (B) · 66 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. | 18.0 s | Very slow |
| Prometheus 2 7B | 67Good quality (B) · 67 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. | 2.2 s | Very slow | 67Good quality (B) · 67 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. | 26.3 s | Very slow |
| Flow Judge v0.1 | 63Good quality (B) · 63 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. | 1.2 s | Slow | 64Good quality (B) · 64 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. | 18.0 s | Very slow |
Estimates come from published hardware specs and our open formulas, not from running the model on your machine. Real speed varies with runtime, drivers and settings. Numbers marked “Measured” come from a linked benchmark.