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 ↓
73731 M4 (10c) 16 GB Qwen3.5-4B 4bit 1k 400.3 38.0 26-03-09
73732 M3 Ultra (80c) 512 GB Qwen3.5-9B 4bit 1k 1,465 105.7 26-03-09
73733 M3 Ultra (80c) 512 GB Qwen3.5-9B 4bit 4k 1,485 101.0 26-03-09
73734 M4 (10c) 16 GB Qwen3.5-9B 4bit 4k 221.7 19.2 26-03-09
73735 M4 (10c) 16 GB Qwen3.5-9B 4bit 1k 221.7 19.4 26-03-09
73736 M5 (10c) 24 GB Qwen3.5-0.8B 8bit 4k 2,382 130.0 26-03-09
73737 M5 (10c) 24 GB Qwen3.5-0.8B 8bit 1k 2,302 132.1 26-03-09
73738 M4 Pro (16c) 24 GB Qwen3.5-9B 4bit 1k 361.4 49.7 26-03-09
73739 M3 (10c) 24 GB Llama-3.2-1B-Instruct 4bit 64k 278.2 8.1 26-03-09
73740 M3 (10c) 24 GB Llama-3.2-1B-Instruct 4bit 32k 596.2 32.2 26-03-09
74,817 results · Page 7374 of 7482