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 ↓ |
|---|---|---|---|---|---|---|---|---|
| 279551 | M4 Pro (20c) | 24 GB | Qwen3-4b-Instruct-2507 | 8bit | 1k | 733.0 | 48.5 | 26-03-19 |
| 279552 | M4 (10c) | 32 GB | Qwen3.5-9B | 4bit | 4k | 221.2 | 19.3 | 26-03-19 |
| 279553 | M4 (10c) | 32 GB | Qwen3.5-9B | 4bit | 1k | 223.3 | 20.7 | 26-03-19 |
| 279554 | M4 (10c) | 16 GB | Qwen3-8B | 4bit | 4k | 205.1 | 10.7 | 26-03-19 |
| 279555 | M3 Max (40c) | 128 GB | Qwen3.5-35B-A3B-Claude-4.... | 8bit | 16k | 1,334 | 52.6 | 26-03-19 |
| 279556 | M3 Max (40c) | 128 GB | Qwen3.5-35B-A3B-Claude-4.... | 8bit | 8k | 1,416 | 64.1 | 26-03-19 |
| 279557 | M3 Max (40c) | 128 GB | Qwen3.5-35B-A3B-Claude-4.... | 8bit | 4k | 1,415 | 68.6 | 26-03-19 |
| 279558 | M3 Max (40c) | 128 GB | Qwen3.5-35B-A3B-Claude-4.... | 8bit | 1k | 1,098 | 71.2 | 26-03-19 |
| 279559 | M4 Pro (20c) | 24 GB | Qwen3-4B-Thinking-2507 | 6bit | 4k | 679.9 | 50.4 | 26-03-19 |
| 279560 | M4 Pro (20c) | 24 GB | Qwen3-4B-Thinking-2507 | 6bit | 1k | 721.6 | 60.8 | 26-03-19 |