← Back to community benchmarks

Qwen3.6-27B-MTPLX-Optimized-Speed

M1 Max (32c) · 32 GB · 4bit · 2026-06-04
Performance
16k
tokens
48.0
PP tok/s
10.2
TG tok/s
341400
TTFT (ms)
15.8
Peak mem (GB)
Hardware
Chip M1 Max (32c)
Memory 32 GB
GPU Cores 32
Software
oMLX v0.4.2.dev1
macOS macOS 26.5
Context 16,384
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 66.2 6.3 15.2 GB view
4k 14.2 10.9 16.2 GB view
16k 48.0 10.2 15.8 GB current
Batching Results
Batch Size TG tok/s Speedup
6.3 1.00×
13.9 2.21×
26.8 4.25×