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 ↓
6501 M3 Ultra (60c) 256 GB Qwen3-Coder-Next 8bit 8k 1,505 58.0 26-03-09
6502 M3 Ultra (60c) 256 GB Qwen3-Coder-Next 8bit 16k 1,423 51.8 26-03-09
6503 M3 Ultra (80c) 512 GB Qwen3.5-27B-heretic 8bit 32k 378.7 17.5 26-03-09
6504 M3 Ultra (80c) 512 GB Qwen3.5-27B-heretic 8bit 16k 403.9 19.2 26-03-09
6505 M2 Max (38c) 32 GB Qwen3.5-35B-A3B 4bit 4k 734.3 69.8 26-03-09
6506 M2 Max (38c) 32 GB Qwen3.5-35B-A3B 4bit 1k 688.7 75.0 26-03-09
6507 M3 Ultra (80c) 512 GB Qwen3.5-27B-heretic 8bit 8k 416.8 20.4 26-03-09
6508 M4 Max (40c) 64 GB Qwen3.5-9B 4bit 1k 858.7 88.6 26-03-09
6509 M3 Ultra (80c) 512 GB Qwen3.5-27B-heretic 8bit 4k 421.8 21.0 26-03-09
6510 M3 Ultra (80c) 512 GB Qwen3.5-27B-heretic 8bit 1k 413.4 21.6 26-03-09
7,562 results · Page 651 of 757