← Back to community benchmarks

Qwen3.6-27B-uncensored-heretic-v2-Native-MTP-Preserved-oQ8

M3 Ultra (80c) · 512 GB · 8bit · 2026-06-21

Performance

8k

tokens

419.0

PP tok/s

21.9

TG tok/s

19553

TTFT (ms)

30.8

Peak mem (GB)

Hardware

Chip M3 Ultra (80c)

Memory 512 GB

GPU Cores 80

Software

oMLX v0.4.4

macOS macOS 26.4.1

Context 8,192

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
4k	418.9	22.3	29.8 GB	view
8k	419.0	21.9	30.8 GB	current
16k	408.0	21.0	32.3 GB	view
32k	382.1	20.2	35.3 GB	view