← Back to community benchmarks

Qwen3.6-27B

M3 Max (40c) · 128 GB · 8bit · 2026-05-19

Performance

32k

tokens

160.9

PP tok/s

7.1

TG tok/s

203641

TTFT (ms)

40.1

Peak mem (GB)

Hardware

Chip M3 Max (40c)

Memory 128 GB

GPU Cores 40

Software

oMLX v0.3.8

macOS macOS 15.7.4

Context 32,768

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	208.4	11.0	33.0 GB	view
4k	210.6	10.9	34.6 GB	view
8k	187.4	10.6	35.6 GB	view
16k	192.1	10.0	37.1 GB	view
32k	160.9	7.1	40.1 GB	current
64k	128.9	6.0	46.2 GB	view
128k	97.0	5.1	58.6 GB	view
195k	79.4	4.5	73.3 GB	view