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 ↓
207611 M3 Max (40c) 128 GB Qwen3.5-35B-A3B-oQ4e 4bit 32k 2,554 34.0 26-04-18
207612 M3 Max (40c) 128 GB Qwen3.5-35B-A3B-oQ4e 4bit 16k 2,044 34.9 26-04-18
207613 M4 (10c) 16 GB gemma-4-E2B-Heretic-Uncen... 4bit 4k 1,814 58.6 26-04-18
207614 M4 (10c) 16 GB gemma-4-E2B-Heretic-Uncen... 4bit 1k 1,386 61.0 26-04-18
207615 M5 Max (40c) 64 GB Qwen3.5-27B-Architect-Pol... 4bit 32k 659.9 25.5 26-04-18
207616 M5 Max (40c) 64 GB Qwen3.5-27B-Architect-Pol... 4bit 16k 770.4 30.6 26-04-18
207617 M5 Max (40c) 64 GB Qwen3.5-27B-Architect-Pol... 4bit 1k 823.0 131.3 26-04-18
207618 M5 Max (40c) 64 GB Qwen3.5-27B-Architect-Pol... 4bit 32k 698.8 26.4 26-04-18
207619 M5 Max (40c) 64 GB Qwen3.5-27B-Architect-Pol... 4bit 16k 768.8 30.7 26-04-18
207620 M5 Max (40c) 64 GB Qwen3.5-27B-Architect-Pol... 4bit 1k 691.5 130.4 26-04-18
318,399 results · Page 20762 of 31840