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 ↓
10981 M2 Ultra (76c) 128 GB Qwen3.5-4B 8bit 32k 1,255 63.0 26-03-09
10982 M2 Ultra (76c) 128 GB Qwen3.5-4B 8bit 8k 1,421 90.1 26-03-09
10983 M2 Ultra (76c) 128 GB Qwen3.5-4B 8bit 16k 1,362 80.1 26-03-09
10984 M2 Ultra (76c) 128 GB Qwen3.5-4B 8bit 4k 1,448 96.2 26-03-09
10985 M2 Ultra (76c) 128 GB Qwen3.5-4B 8bit 1k 1,423 99.9 26-03-09
10986 M4 (10c) 24 GB Llama-3.2-3B-Instruct 4bit 1k 520.4 45.2 26-03-09
10987 M4 (10c) 24 GB Llama-3.2-3B-Instruct 4bit 4k 478.6 37.0 26-03-09
10988 M4 (10c) 24 GB Qwen3.5-4B 4bit 4k 380.6 34.1 26-03-09
10989 M4 (10c) 24 GB Qwen3.5-4B 4bit 1k 378.3 35.1 26-03-09
10990 M2 Ultra (76c) 128 GB Qwen3.5-0.8B bf16 64k 4,660 83.2 26-03-09
11,195 results · Page 1099 of 1120