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 ↓
207711 M3 Ultra (80c) 512 GB Qwen3.6-35B-A3B bf16 4k 2,382 69.1 26-04-19
207712 M3 Max (30c) 96 GB Qwen3.5-4B 4bit 8k 988.3 79.2 26-04-19
207713 M3 Max (30c) 96 GB Qwen3.5-4B 4bit 16k 879.7 68.1 26-04-19
207714 M3 Max (30c) 96 GB Qwen3.5-4B 4bit 4k 994.9 86.2 26-04-19
207715 M3 Max (30c) 96 GB Qwen3.5-4B 4bit 1k 874.5 91.4 26-04-19
207716 M2 Max (38c) 96 GB gemma-4-26b-a4b-it 4bit 16k 622.2 46.2 26-04-19
207717 M2 Max (38c) 96 GB gemma-4-26b-a4b-it 4bit 4k 634.8 62.6 26-04-19
207718 M2 Max (38c) 96 GB gemma-4-26b-a4b-it 4bit 1k 577.3 66.4 26-04-19
207719 M4 Pro (20c) 64 GB gemma-4-31b-it 4bit 1k 100.1 12.2 26-04-19
207720 M4 Pro (20c) 64 GB gemma-4-31b-it 4bit 4k 88.4 5.4 26-04-19
319,084 results · Page 20772 of 31909