Community Benchmarks

Download oMLX →

Real performance numbers from oMLX users around the world. Submit yours from the app (v0.2.6+).

Reset filters
# Chip RAM Model Quant Ctx PP tok/s TG tok/s Date ↓
321 M3 Max (40c) 128 GB Qwen3.5-9B 4bit 64k 586.8 42.9 26-07-20
322 M3 Ultra (80c) 96 GB Ternary-Bonsai-27B 2bit 64k 354.2 38.7 26-07-20
323 M1 Ultra (64c) 128 GB Qwen3.6-35B-A3B 6bit 64k 534.6 46.8 26-07-20
324 M1 Ultra (64c) 128 GB unsloth-gemma-4-31B-it-qa... 4bit 64k 117.5 12.0 26-07-20
325 M5 Max (40c) 128 GB Step-3.7-Flash-oQ3e 3bit 64k 583.1 34.9 26-07-20
326 M3 Max (40c) 128 GB Qwen3.5-9B 4bit 64k 257.5 13.9 26-07-20
327 M5 Max (40c) 128 GB Qwen3.6-35B-A3B 8bit 64k 2,349 68.3 26-07-20
328 M5 Pro (20c) 48 GB Qwen3.5-4B 4bit 64k 1,341 44.2 26-07-20
329 M3 Ultra (60c) 256 GB Ternary-Bonsai-27B 2bit 64k 266.2 32.7 26-07-20
330 M2 Ultra (60c) 128 GB NVIDIA-Nemotron-3-Nano-30... 4bit 64k 878.8 71.8 26-07-20
13,178 results · Page 33 of 1318