← Back to community benchmarks

Qwen3.5-27B4.5bit

M3 (10c) · 24 GB · 4bit · 2026-04-25

Performance

1k

tokens

52.6

PP tok/s

5.9

TG tok/s

19468

TTFT (ms)

15.8

Peak mem (GB)

Hardware

Chip M3 (10c)

Memory 24 GB

GPU Cores 10

Software

oMLX v0.3.6

macOS macOS 26.4

Context 1,024

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	52.6	5.9	15.8 GB	current
4k	53.0	5.8	17.2 GB	view

Batching Results

Batch Size	TG tok/s	Speedup
1×	5.9	1.00×
2×	11.6	1.97×
4×	14.1	2.39×