← Back to community benchmarks

Qwen3.5-4B

M4 (10c) · 16 GB · 4bit · 2026-03-10

Performance

8k

tokens

371.4

PP tok/s

26.7

TG tok/s

22056

TTFT (ms)

4.6

Peak mem (GB)

Hardware

Chip M4 (10c)

Memory 16 GB

GPU Cores 10

Software

oMLX v0.2.6

macOS macOS 26.2

Context 8,192

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	367.9	30.4	3.9 GB	view
4k	376.2	28.9	4.2 GB	view
8k	371.4	26.7	4.6 GB	current

Batching Results

Batch Size	TG tok/s	Speedup
1×	30.4	1.00×
2×	62.1	2.04×
4×	83.7	2.75×