← Back to community benchmarks

Qwen3.5-2B

M5 (10c) · 32 GB · 4bit · 2026-04-18

Performance

128k

tokens

617.3

PP tok/s

22.9

TG tok/s

212315

TTFT (ms)

7.7

Peak mem (GB)

Hardware

Chip M5 (10c)

Memory 32 GB

GPU Cores 10

Software

oMLX v0.3.6

macOS macOS 26.4.1

Context 131,072

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	1,898	70.3	1.9 GB	view
4k	1,536	90.4	2.2 GB	view
8k	1,596	89.2	2.4 GB	view
16k	1,750	76.4	2.7 GB	view
32k	1,549	58.2	3.4 GB	view
64k	1,073	35.6	4.8 GB	view
128k	617.3	22.9	7.7 GB	current
195k	444.8	16.6	10.8 GB	view