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 ↓ |
|---|---|---|---|---|---|---|---|---|
| 155241 | M2 Ultra (76c) | 192 GB | gemma-4-31b-5bit | 5bit | 4k | 179.4 | 18.7 | 26-04-28 |
| 155242 | M2 Ultra (76c) | 192 GB | gemma-4-31b-5bit | 5bit | 1k | 175.9 | 20.3 | 26-04-28 |
| 155243 | M1 Max (32c) | 32 GB | Llama-3.2-3B-Instruct | 8bit | 64k | 275.5 | 13.4 | 26-04-28 |
| 155244 | M1 Max (32c) | 32 GB | Llama-3.2-3B-Instruct | 8bit | 4k | 746.4 | 65.7 | 26-04-28 |
| 155245 | M1 Max (32c) | 32 GB | Llama-3.2-3B-Instruct | 8bit | 1k | 1,134 | 76.7 | 26-04-28 |
| 155246 | M2 Pro (19c) | 32 GB | Qwen3.5-0.8B | 8bit | 4k | 1,973 | 150.0 | 26-04-28 |
| 155247 | M2 Ultra (76c) | 192 GB | Qwen3.6-35B-A3B-5bit | 5bit | 4k | 1,374 | 75.0 | 26-04-28 |
| 155248 | M2 Ultra (76c) | 192 GB | Qwen3.6-35B-A3B-5bit | 5bit | 1k | 1,032 | 78.4 | 26-04-28 |
| 155249 | M5 Max (40c) | 128 GB | Qwen3.6-35B-A3B-Claude-4.... | 4bit | 128k | 1,417 | 42.9 | 26-04-28 |
| 155250 | M4 (10c) | 16 GB | qwen35-9bturboquant-tq3 | 4bit | 4k | 223.0 | 19.7 | 26-04-28 |