← Back to community benchmarks

Qwen3.5-4B

M3 (10c) · 24 GB · 4bit · 2026-03-10

Performance

1k

tokens

340.3

PP tok/s

34.3

TG tok/s

3009

TTFT (ms)

3.9

Peak mem (GB)

Hardware

Chip M3 (10c)

Memory 24 GB

GPU Cores 10

Software

oMLX v0.2.7

macOS macOS 26.3.1

Context 1,024

Batching Results

Batch Size	TG tok/s	Speedup
1×	34.3	1.00×
2×	63.9	1.86×
4×	73.3	2.14×