← Back to community benchmarks

Qwen3.6-27B-oQ8-mtp

M3 Ultra (80c) · 512 GB · 8bit · 2026-05-13

Performance

8k

tokens

418.9

PP tok/s

30.7

TG tok/s

19555

TTFT (ms)

31.3

Peak mem (GB)

Hardware

Chip M3 Ultra (80c)

Memory 512 GB

GPU Cores 80

Software

oMLX v0.3.9.dev2

macOS macOS 26.5

Context 8,192

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	383.2	31.8	28.8 GB	view
4k	418.1	31.4	30.3 GB	view
8k	418.9	30.7	31.3 GB	current

Batching Results

Batch Size	TG tok/s	Speedup
1×	31.8	1.00×
2×	39.2	1.23×
4×	58.1	1.83×