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 ↓ |
|---|---|---|---|---|---|---|---|---|
| 205961 | M3 Ultra (60c) | 256 GB | gemma-4-31b-it | 8bit | 16k | 236.0 | 13.6 | 26-04-15 |
| 205962 | M3 Ultra (60c) | 256 GB | gemma-4-31b-it | 8bit | 4k | 253.4 | 17.6 | 26-04-15 |
| 205963 | M4 Pro (20c) | 24 GB | gemma-4-21b-REAP-Tool-Cal... | 4bit | 1k | 649.7 | 71.1 | 26-04-15 |
| 205964 | M4 Pro (20c) | 24 GB | gemma-4-21b-REAP-Tool-Cal... | 4bit | 4k | 727.7 | 66.8 | 26-04-15 |
| 205965 | M5 (10c) | 16 GB | Qwen3.5-9B-Claude-4.6-Hig... | 8bit | 4k | 637.7 | 14.7 | 26-04-15 |
| 205966 | M5 (10c) | 16 GB | Qwen3.5-9B-Claude-4.6-Hig... | 8bit | 1k | 554.2 | 14.7 | 26-04-15 |
| 205967 | M1 Pro (14c) | 32 GB | Qwen3.5-9B-Claude-4.6-Hig... | 8bit | 4k | 144.3 | 19.0 | 26-04-15 |
| 205968 | M1 Pro (14c) | 32 GB | Qwen3.5-9B-Claude-4.6-Hig... | 8bit | 1k | 143.8 | 19.7 | 26-04-15 |
| 205969 | M3 Max (30c) | 36 GB | Qwen2.5-Coder-14B-Instruc... | 4bit | 4k | 315.9 | 27.2 | 26-04-15 |
| 205970 | M2 (8c) | 16 GB | Qwen3.5-9B | 4bit | 4k | 87.3 | 18.1 | 26-04-15 |