← Back to community benchmarks

Qwen2.5-0.5B-Instruct

M3 Pro (18c) · 36 GB · 4bit · 2026-03-11

Performance

4k

tokens

4,303

PP tok/s

204.8

TG tok/s

952

TTFT (ms)

1.1

Peak mem (GB)

Hardware

Chip M3 Pro (18c)

Memory 36 GB

GPU Cores 18

Software

oMLX v0.2.7

macOS macOS 26.3.1

Context 4,096

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	3,106	251.0	0.8 GB	view
4k	4,303	204.8	1.1 GB	current

Batching Results

Batch Size	TG tok/s	Speedup
1×	251.0	1.00×
2×	434.6	1.73×
4×	540.9	2.15×