← Back to community benchmarks

Qwen3.5-9B-Uncensored-HauhauCS-Aggressivemxfp8

M3 Pro (14c) · 18 GB · 8bit · 2026-05-15

Performance

4k

tokens

261.8

PP tok/s

14.4

TG tok/s

15644

TTFT (ms)

10.3

Peak mem (GB)

Hardware

Chip M3 Pro (14c)

Memory 18 GB

GPU Cores 14

Software

oMLX v0.3.9.dev2

macOS macOS 26.4.1

Context 4,096

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	249.8	15.2	9.7 GB	view
4k	261.8	14.4	10.3 GB	current

Batching Results

Batch Size	TG tok/s	Speedup
1×	15.2	1.00×
2×	26.9	1.77×
4×	27.7	1.82×