Community Benchmarks
Download oMLX →Real performance numbers from oMLX users around the world. Submit yours from the app (v0.2.6+).
| # | Chip | RAM | Model | Quant | Ctx | PP tok/s | TG tok/s | Date ↓ |
|---|---|---|---|---|---|---|---|---|
| 71551 | M3 Ultra (60c) | 96 GB | Qwen3.5-9B | 4bit | 32k | 1,014 | 84.5 | 26-05-18 |
| 71552 | M3 Ultra (60c) | 96 GB | Qwen3.5-9B | 4bit | 64k | 898.2 | 66.9 | 26-05-18 |
| 71553 | M3 Ultra (60c) | 96 GB | Qwen3.5-9B | 4bit | 16k | 1,077 | 97.7 | 26-05-18 |
| 71554 | M3 Ultra (60c) | 96 GB | Qwen3.5-9B | 4bit | 4k | 1,074 | 108.4 | 26-05-18 |
| 71555 | M3 Ultra (60c) | 96 GB | Qwen3.5-9B | 4bit | 8k | 1,108 | 106.5 | 26-05-18 |
| 71556 | M3 Ultra (60c) | 96 GB | Qwen3.5-9B | 4bit | 1k | 907.6 | 110.8 | 26-05-18 |
| 71557 | M5 Max (40c) | 128 GB | gemma-4-E2B-Heretic-Uncen... | 4bit | 1k | 2,160 | 192.5 | 26-05-18 |
| 71558 | M4 (10c) | 16 GB | Qwen3.5-4B | 4bit | 4k | 397.4 | 36.6 | 26-05-18 |
| 71559 | M4 (10c) | 16 GB | Qwen3.5-4B | 4bit | 1k | 383.6 | 38.2 | 26-05-18 |
| 71560 | M2 Pro (16c) | 32 GB | Qwen3-Coder-30B-A3B-Instr... | 4bit | 4k | 276.2 | 37.1 | 26-05-18 |