← Back to community benchmarks

Qwen3.6-27B-oQ4-mtp

M3 Ultra (80c) · 256 GB · 4bit · 2026-06-15

Performance

1k

tokens

393.9

PP tok/s

38.9

TG tok/s

2600

TTFT (ms)

16.4

Peak mem (GB)

Hardware

Chip M3 Ultra (80c)

Memory 256 GB

GPU Cores 80

Software

oMLX v0.4.4rc1

macOS macOS 26.4.1

Context 1,024

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	393.9	38.9	16.4 GB	current
4k	422.5	38.2	17.9 GB	view

Batching Results

Batch Size	TG tok/s	Speedup
1×	38.9	1.00×
2×	66.0	1.70×
4×	82.1	2.11×
8×	94.1	2.42×