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 ↓ |
|---|---|---|---|---|---|---|---|---|
| 335111 | M3 Max (40c) | 128 GB | gpt-oss-20b | unknown | 1k | 724.9 | 74.6 | 26-03-11 |
| 335112 | M3 Max (40c) | 128 GB | gpt-oss-20b | unknown | 4k | 881.8 | 69.3 | 26-03-11 |
| 335113 | M2 Ultra (60c) | 64 GB | Qwen3.5-35B-A3B | 8bit | 4k | 1,125 | 65.7 | 26-03-11 |
| 335114 | M2 Ultra (60c) | 64 GB | Qwen3.5-35B-A3B | 8bit | 1k | 1,075 | 68.3 | 26-03-11 |
| 335115 | M2 (8c) | 16 GB | Llama-3.2-1B-Instruct | 4bit | 4k | 917.4 | 70.2 | 26-03-11 |
| 335116 | M2 (8c) | 16 GB | Llama-3.2-1B-Instruct | 4bit | 1k | 954.0 | 96.8 | 26-03-11 |
| 335117 | M4 (10c) | 32 GB | gpt-oss-20b-MXFP4-Q8 | 4bit | 32k | 251.4 | 21.2 | 26-03-11 |
| 335118 | M3 Pro (14c) | 18 GB | Qwen3.5-4B | 4bit | 4k | 470.7 | 39.7 | 26-03-11 |
| 335119 | M3 Pro (14c) | 18 GB | Qwen3.5-4B | 4bit | 1k | 465.5 | 45.8 | 26-03-11 |
| 335120 | M4 (10c) | 32 GB | gpt-oss-20b-MXFP4-Q8 | 4bit | 4k | 327.7 | 31.1 | 26-03-11 |