modelgrep

Best Local LLMs

Match · Updated July 2026

The best local LLM is MiniMax M3 (44.4 intelligence) — open weights on Hugging Face, small enough to run on your own hardware. DeepSeek V4 Pro (44.3) and Kimi K2.6 (44.2) round out the top three.

44.4Intelligence
$0.300Input /M
1.0MContext

The best local LLMs — open-weight models small enough to run on your own hardware with Ollama, llama.cpp or vLLM — ranked by intelligence. Frontier-scale open models (400B+) are excluded: downloadable isn't the same as runnable. Every model here is on Hugging Face.

  1. 1minimax logo
    minimax-m3
    ReasoningToolsJSON+144.4 intel · $0.300/M · 1.0M ctx
    44.4
    Intelligence
  2. 2deepseek logo
    deepseek-v4-pro
    ReasoningToolsJSON44.3 intel · $0.435/M · 1.0M ctx
    44.3
    Intelligence
  3. 3moonshotai logo
    kimi-k2.6
    ReasoningToolsJSON+144.2 intel · $0.684/M · 262K ctx
    44.2
    Intelligence
  4. 4xiaomi logo
    mimo-v2.5-pro
    ReasoningToolsJSON42.2 intel · $0.435/M · 1.1M ctx
    42.2
    Intelligence
  5. 5moonshotai logo
    kimi-k2.7-code
    ReasoningToolsJSON+141.9 intel · $0.710/M · 262K ctx
    41.9
    Intelligence
  6. 6tencent logo
    hy3
    ReasoningToolsJSON41.2 intel · $0.132/M · 262K ctx
    41.2
    Intelligence
  7. 7N
    nex-n2-pro
    ReasoningToolsVision41.0 intel · $0.250/M · 262K ctx
    41.0
    Intelligence
  8. 8T
    inkling
    ReasoningToolsVision+140.7 intel · $1.00/M · 1.0M ctx
    40.7
    Intelligence
  9. 9deepseek logo
    deepseek-v4-flash
    ReasoningToolsJSON40.3 intel · $0.098/M · 1.0M ctx
    40.3
    Intelligence
  10. 10z-ai logo
    glm-5.1
    ReasoningToolsJSON40.2 intel · $0.966/M · 205K ctx
    40.2
    Intelligence
  11. 11z-ai logo
    glm-5
    ReasoningToolsJSON39.5 intel · $0.950/M · 205K ctx
    39.5
    Intelligence
  12. 12minimax logo
    minimax-m2.7
    ReasoningToolsJSON38.1 intel · $0.250/M · 205K ctx
    38.1
    Intelligence
  13. 13xiaomi logo
    mimo-v2.5
    ReasoningToolsJSON+237.2 intel · $0.140/M · 1.1M ctx
    37.2
    Intelligence
  14. 14qwen logo
    qwen3.6-27b
    ReasoningToolsJSON+137.1 intel · $0.288/M · 262K ctx
    37.1
    Intelligence
  15. 15moonshotai logo
    kimi-k2.5
    ReasoningToolsJSON+135.4 intel · $0.570/M · 262K ctx
    35.4
    Intelligence
  16. 16qwen logo
    qwen3.5-27b
    ReasoningToolsJSON+133.8 intel · $0.195/M · 262K ctx
    33.8
    Intelligence
  17. 17qwen logo
    qwen3.5-397b-a17b
    ReasoningToolsJSON+133.7 intel · $0.390/M · 262K ctx
    33.7
    Intelligence
  18. 18minimax logo
    minimax-m2.5
    ReasoningToolsJSON33.7 intel · $0.150/M · 205K ctx
    33.7
    Intelligence
  19. 19z-ai logo
    glm-4.7
    ReasoningToolsJSON33.7 intel · $0.400/M · 205K ctx
    33.7
    Intelligence
  20. 20tencent logo
    hy3-preview
    ReasoningTools33.6 intel · $0.063/M · 262K ctx
    33.6
    Intelligence
  21. 21M
    longcat-2.0
    ReasoningTools33.5 intel · $0.300/M · 1.0M ctx
    33.5
    Intelligence
  22. 22moonshotai logo
    kimi-k2-thinking
    ReasoningToolsJSON32.7 intel · $0.600/M · 262K ctx
    32.7
    Intelligence
  23. 23qwen logo
    qwen3.5-122b-a10b
    ReasoningToolsJSON+132.3 intel · $0.260/M · 262K ctx
    32.3
    Intelligence
  24. 24qwen logo
    qwen3.6-35b-a3b
    ReasoningToolsJSON+131.6 intel · $0.140/M · 262K ctx
    31.6
    Intelligence
  25. 25minimax logo
    minimax-m2.1
    ReasoningToolsJSON31.4 intel · $0.300/M · 205K ctx
    31.4
    Intelligence

Frequently asked

What is the best local LLM?

The best local LLM is MiniMax M3 (44.4 intelligence) — open weights on Hugging Face, small enough to run on your own hardware. DeepSeek V4 Pro (44.3) and Kimi K2.6 (44.2) round out the top three.

What is the best AI model to run locally?

The best local AI model is MiniMax M3 (44.4 intelligence) — open weights on Hugging Face, small enough to run on your own hardware. DeepSeek V4 Pro (44.3) and Kimi K2.6 (44.2) round out the top three.

What's a good alternative to MiniMax M3?

DeepSeek V4 Pro (44.3) is the closest alternative on this metric, followed by Kimi K2.6 (44.2). See the full ranking above for the tradeoffs.

By hardware

By maker

All rankings