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 ↓
335961 M1 Pro (16c) 16 GB Qwen3.5-4B 4bit 1k 289.3 54.7 26-03-10
335962 M4 (10c) 16 GB Qwen3.5-9B 4bit 4k 216.3 19.0 26-03-10
335963 M4 (10c) 16 GB Qwen3.5-9B 4bit 1k 217.2 19.6 26-03-10
335964 M4 Max (40c) 128 GB NVIDIA-Nemotron-3-Nano-30... 4bit 4k 1,480 149.7 26-03-10
335965 M4 Max (40c) 128 GB NVIDIA-Nemotron-3-Nano-30... 4bit 1k 1,096 147.0 26-03-10
335966 M1 Pro (16c) 16 GB Qwen3.5-2B 6bit 4k 738.5 80.0 26-03-10
335967 M1 Pro (16c) 16 GB Qwen3.5-2B 6bit 1k 715.2 82.5 26-03-10
335968 M1 Max (24c) 32 GB Qwen3.5-9B 4bit 32k 163.8 20.1 26-03-10
335969 M1 Max (24c) 32 GB Qwen3.5-9B 4bit 64k 154.9 14.4 26-03-10
335970 M1 Max (24c) 32 GB Qwen3.5-9B 4bit 16k 177.4 26.7 26-03-10
338,371 results · Page 33597 of 33838