← Back to community benchmarks

Qwen3.6-27B-oQ8-mtp

M5 Max (40c) · 128 GB · 8bit · 2026-05-08
Performance
4k
tokens
694.2
PP tok/s
17.4
TG tok/s
5900
TTFT (ms)
29.8
Peak mem (GB)
Hardware
Chip M5 Max (40c)
Memory 128 GB
GPU Cores 40
Software
oMLX v0.3.9.dev1
macOS macOS 26.4.1
Context 4,096
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 693.1 17.8 28.3 GB view
4k 694.2 17.4 29.8 GB current
8k 634.3 16.9 30.8 GB view
16k 573.5 16.2 32.3 GB view
32k 506.7 15.2 35.3 GB view
64k 337.2 12.3 41.4 GB view
128k 267.2 10.9 53.7 GB view
Batching Results
Batch Size TG tok/s Speedup
17.8 1.00×
26.4 1.48×
33.8 1.90×