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 ↓
2131 M3 Ultra (60c) 96 GB Qwen3.5-0.8B9bit 8bit 64k 4,818 122.4 26-03-09
2132 M3 Ultra (60c) 96 GB Qwen3.5-0.8B9bit 8bit 1k 2,898 301.0 26-03-09
2133 M1 Max (32c) 32 GB Qwen3.5-9B 8bit 64k 220.1 15.6 26-03-09
2134 M1 Max (32c) 32 GB Qwen3.5-9B 8bit 32k 288.6 22.5 26-03-09
2135 M1 Max (32c) 32 GB Qwen3.5-9B 8bit 16k 310.1 28.3 26-03-09
2136 M1 Max (32c) 32 GB Qwen3.5-9B 8bit 8k 316.4 31.9 26-03-09
2137 M1 Max (32c) 32 GB Qwen3.5-9B 8bit 4k 318.6 34.5 26-03-09
2138 M1 Max (32c) 32 GB Qwen3.5-9B 8bit 1k 314.9 35.4 26-03-09
2139 M4 (10c) 16 GB Qwen3.5-VL-4B-4bitCRACK 4bit 4k 373.3 29.3 26-03-09
2140 M4 (10c) 16 GB Qwen3.5-VL-4B-4bitCRACK 4bit 1k 377.5 31.4 26-03-09
2,964 results · Page 214 of 297