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 ↓ |
|---|---|---|---|---|---|---|---|---|
| 324871 | M3 Ultra (80c) | 512 GB | GLM-4.6V-5bit | 5bit | 8k | 591.7 | 29.7 | 26-03-10 |
| 324872 | M3 Ultra (80c) | 512 GB | GLM-4.6V-5bit | 5bit | 4k | 654.1 | 33.8 | 26-03-10 |
| 324873 | M3 Ultra (80c) | 512 GB | GLM-4.6V-5bit | 5bit | 1k | 618.2 | 37.5 | 26-03-10 |
| 324874 | M1 Ultra (64c) | 128 GB | Qwen3.5-27B-heretic | 8bit | 64k | 128.8 | 9.5 | 26-03-10 |
| 324875 | M4 Pro (16c) | 64 GB | Qwen3.5-4B | 8bit | 64k | 427.4 | 26.8 | 26-03-10 |
| 324876 | M4 Pro (16c) | 64 GB | Qwen3.5-4B | 8bit | 1k | 646.9 | 48.4 | 26-03-10 |
| 324877 | M4 Pro (16c) | 64 GB | Qwen3.5-4B | 8bit | 4k | 648.6 | 45.9 | 26-03-10 |
| 324878 | M3 Max (40c) | 128 GB | dolphin-2.9.1-llama-3-70b | 8bit | 4k | 69.0 | 4.8 | 26-03-10 |
| 324879 | M3 Max (40c) | 128 GB | dolphin-2.9.1-llama-3-70b | 8bit | 1k | 83.6 | 4.9 | 26-03-10 |
| 324880 | M2 (10c) | 24 GB | Qwen3.5-9B | 4bit | 1k | 109.5 | 18.2 | 26-03-10 |