← Back to community benchmarks

Qwen2.5-Coder-3B-Instruct

M1 (7c) · 16 GB · 4bit · 2026-04-12

Performance

16k

tokens

73.9

PP tok/s

8.1

TG tok/s

221828

TTFT (ms)

2.7

Peak mem (GB)

Hardware

Chip M1 (7c)

Memory 16 GB

GPU Cores 7

Software

oMLX v0.3.4

macOS macOS 26.3.1

Context 16,384

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
8k	91.6	10.8	2.5 GB	view
16k	73.9	8.1	2.7 GB	current
32k	57.2	10.3	3.3 GB	view