Best local AI for Apple M2 Ultra · 192GB

The best local AI on a Apple M2 Ultra · 192GB right now is DeepSeek-R1 (distill) (70B (Llama) @ Q2_K, ~9–14 tok/s). Full ranking below — capability × real speed on this hardware, never sponsored.

#modelbest fitverdictspeed
1 DeepSeek-R1 (distill) 70B (Llama) Q2_K ✅ Runs comfortably 9–14 tok/s
2 Qwen3 30B-A3B (MoE) Q8_0 ✅ Runs comfortably 60–95 tok/s
3 Qwen2.5 72B Q3_K_M ✅ Runs comfortably 7–11 tok/s
4 Llama 3.1 70B Q3_K_M ✅ Runs comfortably 7–11 tok/s
5 Llama 3.3 70B Q3_K_M ✅ Runs comfortably 7–11 tok/s
6 Mistral Small 3 24B Q4_K_M ✅ Runs comfortably 16–25 tok/s
7 Mixtral 8x7B 8x7B (MoE) Q8_0 ✅ Runs comfortably 17–30 tok/s
8 Gemma 2 27B Q3_K_M ✅ Runs comfortably 15–25 tok/s
9 Gemma 4 12B Q8_0 ✅ Runs comfortably 15–25 tok/s
10 Gemma 3 27B Q8_0 ✅ Runs comfortably 8–13 tok/s
11 Phi-4 14B Q5_K_M ✅ Runs comfortably 20–35 tok/s
12 Mistral 7B 7B Q5_K_M ✅ Runs comfortably 40–70 tok/s
$npx runlocal-sh install deepseek-r1

Rankings use the same GQA-aware fit + bandwidth-calibrated speed math as the CLI — no sponsorships, no affiliate re-sorting, ever.

Different machine? Apple M1 · 8GB · Apple M1 Pro · 16GB · Apple M2 · 16GB · Apple M3 Pro · 18GB · Apple M2 Pro · 32GB · Apple M4 Pro · 48GB

Or check a specific model on your exact rig: open the checker →