Best local LLMs for the Rockchip RK3588 / RK3588S (16 GB)

Each list assumes the context that task needs, so a model that fits for chat may rank lower for long documents.

General chat at 8K context

  1. 77Great quality (A) · 77 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 84/100 (quality, fit, speed and memory use) sets the ranking. K2 Horizon 7B Q5_K_M · 3.2 tok/s · Slow
  2. 64Good quality (B) · 64 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. Gemma 4 26B-A4B Q2_K · 14.3 tok/s · Runs well (tight fit)
  3. 74Great quality (A) · 74 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. SenseNova Loop 9B Preview Q4_K_M · 3.6 tok/s · Slow
  4. 73Great quality (A) · 73 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. Ornith 1.5 9B Q4_K_M · 3.6 tok/s · Slow
  5. 71Great quality (A) · 71 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. TMax 9B Q5_K_M · 3.2 tok/s · Slow

Coding at 32K context

  1. 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. Mellum2.1 Thinking (12B-A2.5B) Q4_K_M · 13.2 tok/s · Runs well
  2. 75Great quality (A) · 75 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. MiMo V2.6 Distill Qwen 9B Q5_K_M · 3.0 tok/s · Slow
  3. 77Great quality (A) · 77 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. K2 Horizon 7B Q5_K_M · 3.2 tok/s · Slow (tight fit)
  4. 58Good quality (B) · 58 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. Maple Preview (20B-A1B) Q3_K_M · 27.3 tok/s · Runs well (tight fit)
  5. 74Great quality (A) · 74 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. SenseNova Loop 9B Preview Q4_K_M · 3.6 tok/s · Slow

Reasoning at 32K context

  1. 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. Mellum2.1 Thinking (12B-A2.5B) Q4_K_M · 13.2 tok/s · Runs well
  2. 75Great quality (A) · 75 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. MiMo V2.6 Distill Qwen 9B Q5_K_M · 3.0 tok/s · Slow
  3. 77Great quality (A) · 77 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. K2 Horizon 7B Q5_K_M · 3.2 tok/s · Slow (tight fit)
  4. 58Good quality (B) · 58 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. Maple Preview (20B-A1B) Q3_K_M · 27.3 tok/s · Runs well (tight fit)
  5. 74Great quality (A) · 74 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. SenseNova Loop 9B Preview Q4_K_M · 3.6 tok/s · Slow

Documents (RAG) at 32K context

  1. 74Great quality (A) · 74 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. SenseNova Loop 9B Preview Q4_K_M · 3.6 tok/s · Slow
  2. 73Great quality (A) · 73 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. Ornith 1.5 9B Q4_K_M · 3.6 tok/s · Slow
  3. 72Great quality (A) · 72 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. Qwen3.5 9B Q4_K_M · 3.6 tok/s · Slow
  4. 71Great quality (A) · 71 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. TMax 9B Q5_K_M · 3.2 tok/s · Slow
  5. 71Great quality (A) · 71 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. Ornith 1.0 9B Q4_K_M · 3.6 tok/s · Slow

Agents & tools at 64K context

  1. 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. Mellum2.1 Thinking (12B-A2.5B) Q4_K_M · 13.2 tok/s · Runs well
  2. 75Great quality (A) · 75 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. MiMo V2.6 Distill Qwen 9B Q5_K_M · 3.0 tok/s · Slow
  3. 74Great quality (A) · 74 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. SenseNova Loop 9B Preview Q4_K_M · 3.6 tok/s · Slow
  4. 69Good quality (B) · 69 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. Gemma 4 12B Q3_K_M · 3.7 tok/s · Slow (tight fit)
  5. 67Good quality (B) · 67 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking. UI-Mate 9B Q5_K_M · 3.0 tok/s · Slow

Reverse engineering at 32K context

  1. 66Good quality (B) · 66 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. Mellum2.1 Thinking (12B-A2.5B) Q4_K_M · 13.2 tok/s · Runs well
  2. 75Great quality (A) · 75 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. MiMo V2.6 Distill Qwen 9B Q5_K_M · 3.0 tok/s · Slow
  3. 77Great quality (A) · 77 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking. K2 Horizon 7B Q5_K_M · 3.2 tok/s · Slow (tight fit)
  4. 58Good quality (B) · 58 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. Maple Preview (20B-A1B) Q3_K_M · 27.3 tok/s · Runs well (tight fit)
  5. 74Great quality (A) · 74 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. SenseNova Loop 9B Preview Q4_K_M · 3.6 tok/s · 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.

Was this page helpful?