← Back to community benchmarks

Qwen3.6-35B-oQ8e-mtp

M3 Ultra (80c) · 512 GB · 8bit · 2026-07-13

Performance

16k

tokens

2,619

PP tok/s

71.5

TG tok/s

6256

TTFT (ms)

36.8

Peak mem (GB)

Hardware

Chip M3 Ultra (80c)

Memory 512 GB

GPU Cores 80

Software

oMLX v0.5.1

macOS macOS 26.5.2

Context 16,384

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	2,640	76.6	35.4 GB	view
4k	3,022	75.4	36.1 GB	view
8k	2,884	73.7	36.4 GB	view
16k	2,619	71.5	36.8 GB	current
32k	2,218	67.4	37.4 GB	view
64k	1,690	59.2	38.6 GB	view
128k	1,137	50.2	40.1 GB	view