Local LLMs

LLMs you can run locally, with memory needs per quantization, context length and license.

Step 1

Tell us your hardware first

Then you will see how many of these 621 LLMs run on it, and can filter the list by how well each one runs.

Detect my hardware Pick my GPU or device

Model	Made by	Arch	Parameters	Max context	Quality tier	Released	Runs in
Kimi K3.1 Pro (2.8T)	Moonshot AI	MoE	2,780B (104.0B active)	1,024K	99.3	2026-10 New
DeepSeek V4.1 Flash 1017 (552B-A16B)	DeepSeek	MoE	763B (16.0B active)	1,024K	98.3	2026-10	+14
GLM-5.3 (753B-A40B)	Zhipu AI	MoE	753B (40.0B active)	1,024K	98.3	2026-08	+17
Kimi K3 (2.8T)	Moonshot AI	MoE	2,780B (104.0B active)	1,024K	98	2026-07	+14
Qwen3.8 Max (2.4T-A95B)	Alibaba	MoE	2,446B (95.0B active)	256K	98.3	2026-08	+14
Ling 3.1 Flash 1019 (560B-A25B)	Ant Group (inclusionAI)	MoE	572B (25.0B active)	1,024K	97.3	2026-10	+14
LongCat 2.5 (1.6T-A48B)	Meituan	MoE	1,780B (48.0B active)	512K	97.7	2026-11 New
MiMo V2.6 Pro (1T-A42B)	Xiaomi	MoE	1,024B (42.0B active)	1,024K	97.4	2026-09	+17
MiniMax M3.1 Flash (428B-A23B)	MiniMax	MoE	429B (23.0B active)	1,024K	97	2026-10	+14
Nex N2.6 (321B-A18B)	Nex-AGI	MoE	321B (18.0B active)	1,024K	97	2026-11 New	+14
Step 5 Preview (600B-A27B)	StepFun	MoE	604B (27.0B active)	1,024K	97	2026-09
DeepSeek V4 Pro (1.6T-A49B)	DeepSeek	MoE	1,651B (49.0B active)	1,024K	96	2026-08	+14
DeepSeek V4.1 Flash (552B-A16B)	DeepSeek	MoE	763B (16.0B active)	1,024K	96.5	2026-09
GLM-5.2 (753B-A40B)	Zhipu AI	MoE	753B (40.0B active)	1,024K	96.3	2026-06	+17
GLM-5.3 Flash (320B-A18B)	Zhipu AI	MoE	321B (18.0B active)	1,024K	96.9	2026-08	+17
Hunyuan Hy4 Preview (780B)	Tencent	MoE	780B (55.0B active)	1,024K	96.3	2026-08	+17
MiMo V2.6 Flash (310B-A15B)	Xiaomi	MoE	311B (15.0B active)	1,024K	96.6	2026-09	+17
Mistral Large 4 (1T-A52B)	Mistral AI	MoE	1,053B (52.0B active)	1,024K	96.5	2026-10	+14
Muse Spark Open (816B-A42B)	Meta	MoE	822B (42.0B active)	1,024K	96.3	2026-10	+17
Qwen3.8 Flash Next (125B-A6B)	Alibaba	MoE	180B (6.0B active)	256K	96.5	2026-08	+17
DeepSeek V4 Pro (April 2026)	DeepSeek	MoE	1,599B (49.0B active)	1,024K	95	2026-04	+6
GLM-5.1 (744B-A40B)	Zhipu AI	MoE	754B (40.0B active)	198K	95.2	2026-04	+17
Intern-S2 (397B-A17B)	Shanghai AI Lab	MoE	403B (17.0B active)	256K	95.3	2026-09	+14
Kimi K2.6 (1T)	Moonshot AI	MoE	1,027B (32.0B active)	256K	95.7	2026-04	+14
Ling 3.1 Flash (560B-A25B)	Ant Group (inclusionAI)	MoE	572B (25.0B active)	1,024K	95.5	2026-10	+14
MiMo V2.5 Pro (1T-A42B)	Xiaomi	MoE	1,023B (42.0B active)	1,024K	95.5	2026-04	+14
Naive N0.5 Flash (309B-A15.5B)	Naive AI	MoE	309B (15.5B active)	1,024K	95	2026-09
Nanbeige4.3 33B	BOSS Zhipin (Nanbeige)	Dense	33B	512K	95.6	2026-10	+14
Nex N2 Pro (397B-A17B)	Nex-AGI	MoE	397B (17.0B active)	256K	95.2	2026-06	+17
Qwen3.5 397B-A17B	Alibaba	MoE	403B (17.0B active)	256K	95	2026-02	+17
Qwen3.8 27B	Alibaba	Dense	28B	256K	95.5	2026-08	+17
Beam 1.0 (501B-A23B)	Reflection AI	MoE	501B (23.4B active)	1,024K	94.3	2026-10	+14
DeepSeek V4 Flash (284B-A13B)	DeepSeek	MoE	304B (13.0B active)	1,024K	94	2026-07	+14
GLM-5 (744B-A40B)	Zhipu AI	MoE	754B (40.0B active)	198K	94.7	2026-02	+17
Hunyuan Hy3 (295B-A21B)	Tencent	MoE	299B (21.0B active)	256K	94.4	2026-07	+14
K2-Horizon (375B-A23B)	Institute of Foundation Models	MoE	379B (23.0B active)	512K	94.5	2026-09	+14
Kimi K2.5 (1T)	Moonshot AI	MoE	1,027B (32.0B active)	256K	94.3	2026-01	+17
Kimi K2.7 Code (1T)	Moonshot AI	MoE	1,027B (32.0B active)	256K	94.3	2026-06	+17
MiMo V2.5 (310B-A15B)	Xiaomi	MoE	311B (15.0B active)	1,024K	94	2026-04	+14
MiniMax M3 (427B)	MiniMax	MoE	427B (25.0B active)	1,024K	94.8	2026-06	+17
Ornith 1.5 397B-A17B	DeepReinforce	MoE	403B (17.0B active)	256K	94	2026-08	+14
Qwen3.6 27B	Alibaba	Dense	28B	256K	94	2026-04	+17
DeepSeek R1 0528 (671B)	DeepSeek	MoE	685B (37.0B active)	160K	93	2025-05	+18
DeepSeek V3.2 (671B)	DeepSeek	MoE	685B (37.0B active)	160K	93.7	2025-12	+17
GLM-4.7 (355B-A32B)	Zhipu AI	MoE	358B (32.0B active)	198K	93.5	2025-12	+17
Inkling (975B-A41B)	Thinking Machines Lab	MoE	952B (41.0B active)	1,024K	93.5	2026-07	+17
Inkling Small (276B-A12B)	Thinking Machines Lab	MoE	266B (12.0B active)	1,024K	93.7	2026-07	+17
Kimi K2 Thinking (1T)	Moonshot AI	MoE	1,026B (32.0B active)	256K	93	2025-11	+17
MiniCPM-V 4.7 35B-A3B	OpenBMB	MoE	35B (3.0B active)	256K	93.3	2026-10	+17
MiniMax M2.7 (230B-A10B)	MiniMax	MoE	229B (10.0B active)	200K	93.5	2026-04	+17

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.