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 ↓ |
|---|---|---|---|---|---|---|---|---|
| 31 | M4 (10c) | 32 GB | Qwen3.5-9B | 4bit | 4k | 209.8 | 18.4 | 26-03-09 |
| 32 | M4 (10c) | 32 GB | Qwen3.5-9B | 4bit | 1k | 210.3 | 19.6 | 26-03-09 |
| 33 | M2 Ultra (60c) | 64 GB | Qwen3.5-4B | 4bit | 32k | 1,002 | 76.4 | 26-03-09 |
| 34 | M2 Ultra (60c) | 64 GB | Qwen3.5-4B | 4bit | 64k | 855.5 | 50.5 | 26-03-09 |
| 35 | M2 Ultra (60c) | 64 GB | Qwen3.5-4B | 4bit | 8k | 1,132 | 113.4 | 26-03-09 |
| 36 | M2 Ultra (60c) | 64 GB | Qwen3.5-4B | 4bit | 16k | 1,087 | 98.1 | 26-03-09 |
| 37 | M2 Ultra (60c) | 64 GB | Qwen3.5-4B | 4bit | 1k | 969.4 | 129.0 | 26-03-09 |
| 38 | M2 Ultra (60c) | 64 GB | Qwen3.5-4B | 4bit | 4k | 1,112 | 123.1 | 26-03-09 |
| 39 | M3 Max (40c) | 128 GB | Qwen3-Coder-Next | 8bit | 4k | 1,017 | 48.7 | 26-03-09 |
| 40 | M3 Max (40c) | 128 GB | Qwen3-Coder-Next | 8bit | 1k | 837.2 | 51.9 | 26-03-09 |