← Back to community benchmarks

Qwen3.5-4B

M3 Max (30c) · 36 GB · 4bit · 2026-05-03

Performance

4k

tokens

979.4

PP tok/s

87.4

TG tok/s

4182

TTFT (ms)

4.1

Peak mem (GB)

Hardware

Chip M3 Max (30c)

Memory 36 GB

GPU Cores 30

Software

oMLX v0.3.8

macOS macOS 26.4.1

Context 4,096

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	862.0	91.2	3.4 GB	view
4k	979.4	87.4	4.1 GB	current

Batching Results

Batch Size	TG tok/s	Speedup
1×	91.2	1.00×
2×	164.2	1.80×