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 ↓
206971 M3 Ultra (80c) 256 GB Qwen3-Coder-Next 8bit 1k 1,576 66.8 26-04-17
206972 M3 Ultra (80c) 256 GB Qwen3-Coder-Next 8bit 4k 2,058 65.3 26-04-17
206973 M4 Pro (20c) 48 GB gemma-4-26b-a4b-it-UD 4bit 4k 737.0 52.8 26-04-17
206974 M4 Pro (20c) 48 GB gemma-4-26b-a4b-it-UD 4bit 1k 667.5 57.1 26-04-17
206975 M5 Pro (16c) 48 GB Qwen3-4B-Instruct-2507 4bit 4k 2,235 76.4 26-04-17
206976 M5 Pro (16c) 48 GB Qwen3-4B-Instruct-2507 4bit 1k 1,992 95.3 26-04-17
206977 M5 Pro (16c) 48 GB Qwen3.5-4B-Neo 4bit 4k 2,118 91.5 26-04-17
206978 M5 Pro (16c) 48 GB Qwen3.5-4B-Neo 4bit 1k 1,647 96.8 26-04-17
206979 M2 Pro (16c) 32 GB gemma-4-31B-it-TurboQuant 4bit 8k 34.8 2.1 26-04-17
206980 M2 Pro (16c) 32 GB gemma-4-31B-it-TurboQuant 4bit 4k 40.0 6.9 26-04-17
312,437 results · Page 20698 of 31244