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 ↓ |
|---|---|---|---|---|---|---|---|---|
| 337731 | M3 Ultra (60c) | 256 GB | Qwen3.5-27B | 4bit | 8k | 318.8 | 33.5 | 26-03-12 |
| 337732 | M3 Ultra (60c) | 256 GB | Qwen3.5-27B | 4bit | 16k | 310.2 | 30.2 | 26-03-12 |
| 337733 | M3 Ultra (60c) | 256 GB | Qwen3.5-27B | 4bit | 1k | 320.7 | 37.7 | 26-03-12 |
| 337734 | M3 Ultra (60c) | 256 GB | Qwen3.5-27B | 4bit | 4k | 324.7 | 35.6 | 26-03-12 |
| 337735 | M4 (10c) | 16 GB | Qwen3.5-9B | 4bit | 4k | 224.9 | 20.0 | 26-03-12 |
| 337736 | M4 (10c) | 16 GB | Qwen3.5-9B | 4bit | 1k | 220.0 | 20.8 | 26-03-12 |
| 337737 | M4 (10c) | 16 GB | Qwen3.5-9B | 4bit | 1k | 225.6 | 21.3 | 26-03-12 |
| 337738 | M4 (10c) | 16 GB | Qwen3.5-9B | 4bit | 4k | 224.5 | 20.2 | 26-03-12 |
| 337739 | M4 (10c) | 16 GB | Qwen3.5-9B | 4bit | 1k | 224.2 | 20.7 | 26-03-12 |
| 337740 | M2 Max (38c) | 32 GB | Llama-3.2-3B-Instruct | 4bit | 4k | 1,289 | 103.8 | 26-03-12 |