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 ↓
207671 M1 Pro (14c) 16 GB Qwen3.5-9B 4bit 16k 133.9 22.9 26-04-19
207672 M3 Ultra (60c) 256 GB Qwen3.6-35B-A3B 8bit 32k 1,739 62.6 26-04-19
207673 M3 Ultra (60c) 256 GB Qwen3.6-35B-A3B 8bit 8k 2,113 79.0 26-04-19
207674 M3 Ultra (60c) 256 GB Qwen3.6-35B-A3B 8bit 1k 1,489 86.0 26-04-19
207675 M1 Pro (14c) 16 GB Qwen3.5-9B 4bit 16k 131.6 18.1 26-04-19
207676 M5 Max (40c) 128 GB Qwen3.6-35B-A3B 8bit 4k 3,682 97.8 26-04-19
207677 M5 Max (40c) 128 GB Qwen3.6-35B-A3B 8bit 1k 2,203 101.5 26-04-19
207678 M3 Ultra (60c) 256 GB Qwen3.6-35B-A3B-oQ8 8bit 32k 1,734 61.3 26-04-19
207679 M3 Ultra (60c) 256 GB Qwen3.6-35B-A3B-oQ8 8bit 8k 2,114 75.0 26-04-19
207680 M3 Ultra (60c) 256 GB Qwen3.6-35B-A3B-oQ8 8bit 1k 1,488 85.1 26-04-19
318,836 results · Page 20768 of 31884