← Back to community benchmarks

Qwen3.6-27B-oQ5-mtp

M3 Max (40c) · 128 GB · 5bit · 2026-05-22
Performance
4k
tokens
161.1
PP tok/s
15.3
TG tok/s
25418
TTFT (ms)
21.2
Peak mem (GB)
Hardware
Chip M3 Max (40c)
Memory 128 GB
GPU Cores 40
Software
oMLX v0.3.9
macOS macOS 26.3.1
Context 4,096
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 151.0 13.6 19.8 GB view
4k 161.1 15.3 21.2 GB current
8k 170.1 16.1 22.3 GB view
16k 156.3 9.9 23.8 GB view
Batching Results
Batch Size TG tok/s Speedup
13.6 1.00×
9.7 0.71×
11.7 0.86×