← Back to community benchmarks

Qwen36-27B-oQ4-mtp

M4 Max (40c) · 128 GB · 4bit · 2026-05-22
Performance
16k
tokens
239.0
PP tok/s
34.2
TG tok/s
68557
TTFT (ms)
20.7
Peak mem (GB)
Hardware
Chip M4 Max (40c)
Memory 128 GB
GPU Cores 40
Software
oMLX v0.3.9
macOS macOS 15.7.4
Context 16,384
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 239.9 40.6 16.7 GB view
4k 244.9 39.1 18.2 GB view
8k 246.2 37.4 19.2 GB view
16k 239.0 34.2 20.7 GB current
32k 226.0 32.4 23.7 GB view
64k 200.0 25.5 29.7 GB view