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 ↓ |
|---|---|---|---|---|---|---|---|---|
| 156701 | M3 Ultra (60c) | 96 GB | gemma-4-31b-it | 4bit | 4k | 255.5 | 29.7 | 26-05-04 |
| 156702 | M5 Max (40c) | 128 GB | Qwen3.6-35B-A3B | 8bit | 4k | 3,283 | 98.8 | 26-05-04 |
| 156703 | M5 Max (40c) | 128 GB | Qwen3.6-35B-A3B | 8bit | 1k | 1,592 | 98.8 | 26-05-04 |
| 156704 | M3 Max (40c) | 64 GB | KyleHessling1-Qwopus-GLM-... | 4bit | 4k | 374.0 | 36.4 | 26-05-04 |
| 156705 | M3 Max (40c) | 64 GB | KyleHessling1-Qwopus-GLM-... | 4bit | 1k | 354.8 | 37.3 | 26-05-04 |
| 156706 | M4 Pro (20c) | 64 GB | gemma-4-26B-A4B-it | 6bit | 4k | 691.9 | 51.5 | 26-05-04 |
| 156707 | M4 Pro (20c) | 64 GB | gemma-4-26B-A4B-it | 6bit | 8k | 663.7 | 48.4 | 26-05-04 |
| 156708 | M5 Max (40c) | 128 GB | Qwen3-4B-Instruct-2507 | bf16 | 1k | 3,672 | 56.9 | 26-05-04 |
| 156709 | M4 Pro (20c) | 64 GB | Qwen3.6-35B-A3B-Heretic | 4bit | 1k | 704.0 | 206.0 | 26-05-04 |
| 156710 | M4 Pro (20c) | 64 GB | Qwen3.6-35B-A3B-Heretic | 4bit | 4k | 584.8 | 78.5 | 26-05-04 |