| quant | needs | speed |
|---|---|---|
| ✓ Q8_0 | 46.3 GB | ~5–9 tok/s (slow) |
| ✓ Q5_K_M | 34.3 GB | ~7–12 tok/s (usable) |
| ✓ Q4_K_M | 30.7 GB | ~8–13 tok/s (usable) |
| ✓ Q3_K_M | 26.7 GB | ~9–15 tok/s (usable) |
Same model, other machines: Apple M1 · Apple M1 Pro · Apple M2 · Apple M3 Pro · Apple M2 Pro · Apple M4 Pro
Also runs on a Apple M2 Ultra: DeepSeek-R1 (distill) · Qwen3 · Qwen2.5 · Llama 3.1 · Llama 3.3
Best models for a Apple M2 Ultra · 192GB →
Check any combo yourself: open the checker →