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 ↓
6581 M1 Max (32c) 64 GB Qwen3.5-9B 8bit 1k 316.3 35.9 26-03-09
6582 M1 Max (32c) 64 GB Qwen3.5-9B 8bit 4k 321.1 34.7 26-03-09
6583 M1 Max (32c) 64 GB Qwen3.5-35B-A3B 6bit 1k 524.7 50.4 26-03-09
6584 M1 Max (32c) 64 GB Qwen3.5-35B-A3B 6bit 4k 558.7 49.0 26-03-09
6585 M1 Max (32c) 64 GB Qwen3.5-35B-A3B 4bit 1k 540.2 63.7 26-03-09
6586 M1 Max (32c) 64 GB Qwen3.5-35B-A3B 4bit 4k 585.0 60.1 26-03-09
6587 M1 Max (32c) 64 GB Qwen3.5-9B 8bit 1k 304.3 35.6 26-03-09
6588 M1 Max (32c) 64 GB Qwen3.5-9B 8bit 4k 319.1 34.7 26-03-09
6589 M3 Ultra (80c) 512 GB GLM-5-4.8bit 4bit 32k 73.2 12.3 26-03-09
6590 M3 Ultra (80c) 512 GB GLM-5-4.8bit 4bit 16k 119.8 13.0 26-03-09
6,770 results · Page 659 of 677