← Back to community benchmarks

Qwen3.6-35B-A3B-oQ4

M3 Ultra (80c) · 512 GB · 4bit · 2026-05-10

Performance

32k

tokens

2,075

PP tok/s

72.5

TG tok/s

15792

TTFT (ms)

23.0

Peak mem (GB)

Hardware

Chip M3 Ultra (80c)

Memory 512 GB

GPU Cores 80

Software

oMLX v0.3.9.dev1

macOS macOS 15.3

Context 32,768

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	1,647	87.4	19.9 GB	view
4k	2,420	83.3	20.7 GB	view
32k	2,075	72.5	23.0 GB	current

Batching Results

Batch Size	TG tok/s	Speedup
1×	87.4	1.00×
2×	144.1	1.65×
4×	222.4	2.54×