← Back to community benchmarks

Qwen2.5-7B-Instruct

M5 (10c) · 32 GB · 4bit · 2026-03-11

Performance

8k

tokens

89.3

PP tok/s

2.7

TG tok/s

91776

TTFT (ms)

5.0

Peak mem (GB)

Hardware

Chip M5 (10c)

Memory 32 GB

GPU Cores 10

Software

oMLX v0.2.7

macOS macOS 26.4

Context 8,192

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
4k	64.6	6.3	4.8 GB	view
8k	89.3	2.7	5.0 GB	current
16k	174.6	21.5	5.5 GB	view
32k	168.9	2.3	6.5 GB	view