← Back to community benchmarks

Qwen3.6-27B-MTP-4bit

M5 (10c) · 24 GB · 4bit · 2026-05-22

Performance

1k

tokens

196.1

PP tok/s

13.4

TG tok/s

5221

TTFT (ms)

16.0

Peak mem (GB)

Hardware

Chip M5 (10c)

Memory 24 GB

GPU Cores 10

Software

oMLX v0.3.9

macOS macOS 26.5

Context 1,024

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	196.1	13.4	16.0 GB	current
4k	203.3	14.2	17.4 GB	view