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 ↓ |
|---|---|---|---|---|---|---|---|---|
| 71001 | M3 Pro (18c) | 36 GB | gemma-4-26B-A4B-RotorQuan... | 4bit | 8k | 553.1 | 33.6 | 26-05-20 |
| 71002 | M3 Pro (18c) | 36 GB | gemma-4-26B-A4B-RotorQuan... | 4bit | 4k | 563.9 | 39.8 | 26-05-20 |
| 71003 | M3 Pro (18c) | 36 GB | gemma-4-26B-A4B-RotorQuan... | 4bit | 1k | 523.2 | 41.4 | 26-05-20 |
| 71004 | M3 Ultra (60c) | 256 GB | Llama-4-Scout-17B-16E-Ins... | 6bit | 4k | 496.1 | 30.0 | 26-05-20 |
| 71005 | M3 Ultra (60c) | 256 GB | Llama-4-Scout-17B-16E-Ins... | 6bit | 1k | 454.5 | 32.0 | 26-05-20 |
| 71006 | M5 Pro (20c) | 48 GB | Qwen3.6-35B-A3B | 4bit | 4k | 2,103 | 94.9 | 26-05-20 |
| 71007 | M5 Pro (20c) | 48 GB | Qwen3.6-35B-A3B | 4bit | 1k | 1,139 | 97.7 | 26-05-20 |
| 71008 | M2 Max (30c) | 32 GB | Qwen3.5-9B-oQ4-mtp | 4bit | 32k | 243.8 | 31.4 | 26-05-20 |
| 71009 | M2 Max (30c) | 64 GB | Qwen3.6-27B-oQ4 | 4bit | 16k | 152.4 | 18.2 | 26-05-20 |
| 71010 | M2 Max (30c) | 64 GB | Qwen3.6-27B-oQ4 | 4bit | 1k | 153.9 | 20.5 | 26-05-20 |