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 ↓
337281 M4 Max (40c) 128 GB Qwen3-Coder-Next 8bit 32k 897.7 42.6 26-03-10
337282 M4 Max (40c) 128 GB Qwen3-Coder-Next 8bit 16k 1,059 48.3 26-03-10
337283 M4 Max (40c) 128 GB Qwen3-Coder-Next 8bit 8k 1,176 57.2 26-03-10
337284 M4 Max (40c) 128 GB Qwen3-Coder-Next 8bit 1k 902.1 60.6 26-03-10
337285 M4 Max (40c) 128 GB Qwen3-Coder-Next 8bit 4k 1,193 59.7 26-03-10
337286 M4 (10c) 32 GB Qwen3.5-9B-OptiQ 4bit 1k 214.4 19.7 26-03-10
337287 M4 (10c) 16 GB Qwen3.5-4B 4bit 1k 406.8 37.6 26-03-10
337288 M4 (10c) 16 GB Qwen3.5-4B 4bit 4k 404.4 37.2 26-03-10
337289 M2 Max (30c) 64 GB Llama-3.3-70B-Instruct 4bit 4k 54.3 7.9 26-03-10
337290 M2 Max (30c) 64 GB Llama-3.3-70B-Instruct 4bit 1k 58.3 8.5 26-03-10
342,367 results · Page 33729 of 34237