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 ↓
155891 M3 Ultra (80c) 512 GB Qwen3-Coder-Next 8bit 4k 1,993 58.3 26-04-30
155892 M4 (10c) 24 GB gemma-4-e4b-it 4bit 4k 615.8 32.3 26-04-30
155893 M4 (10c) 24 GB gemma-4-e4b-it 4bit 1k 642.5 34.0 26-04-30
155894 M3 Ultra (80c) 512 GB Qwen3.5-122B-A10B 4bit 1k 749.4 56.2 26-04-30
155895 M3 Ultra (80c) 512 GB Qwen3.5-122B-A10B 4bit 4k 939.4 52.8 26-04-30
155896 M5 Max (40c) 128 GB Qwen3.6-27B-UD 4bit 8k 855.1 19.3 26-04-30
155897 M5 Max (40c) 128 GB Qwen3.6-27B-UD 4bit 16k 831.5 19.8 26-04-30
155898 M5 Max (40c) 128 GB Qwen3.6-27B-UD 4bit 1k 792.2 21.4 26-04-30
155899 M5 Max (40c) 128 GB Qwen3.6-27B-UD 4bit 4k 930.4 21.0 26-04-30
155900 M5 Max (40c) 128 GB Qwen3.6-27B-UD 4bit 16k 790.4 19.7 26-04-30
311,551 results · Page 15590 of 31156