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 ↓ |
|---|---|---|---|---|---|---|---|---|
| 329881 | M3 Ultra (60c) | 256 GB | Qwen3.5-35B-A3B-5bit | 5bit | 16k | 1,887 | 73.6 | 26-03-10 |
| 329882 | M3 Ultra (60c) | 256 GB | Qwen3.5-35B-A3B-5bit | 5bit | 32k | 1,636 | 64.0 | 26-03-10 |
| 329883 | M3 Ultra (60c) | 256 GB | Qwen3.5-35B-A3B-5bit | 5bit | 4k | 2,093 | 85.0 | 26-03-10 |
| 329884 | M3 Ultra (60c) | 256 GB | Qwen3.5-35B-A3B-5bit | 5bit | 8k | 2,030 | 80.4 | 26-03-10 |
| 329885 | M3 Ultra (60c) | 256 GB | Qwen3.5-35B-A3B-5bit | 5bit | 1k | 2,010 | 87.8 | 26-03-10 |
| 329886 | M2 Max (38c) | 64 GB | Qwen3.5-4B | 8bit | 1k | 723.9 | 62.3 | 26-03-10 |
| 329887 | M2 Max (38c) | 64 GB | Qwen3.5-4B | 8bit | 4k | 738.6 | 59.8 | 26-03-10 |
| 329888 | M1 Max (24c) | 32 GB | Qwen3.5-9B | 4bit | 64k | 205.6 | 17.0 | 26-03-10 |
| 329889 | M1 Max (24c) | 32 GB | Qwen3.5-9B | 4bit | 32k | 222.7 | 28.8 | 26-03-10 |
| 329890 | M1 Max (24c) | 32 GB | Qwen3.5-9B | 4bit | 8k | 238.2 | 41.4 | 26-03-10 |