← Back to community benchmarks

Qwen3.5-9B

M3 Ultra (80c) · 512 GB · 4bit · 2026-03-16

Performance

4k

tokens

1,491

PP tok/s

100.6

TG tok/s

2747

TTFT (ms)

7.0

Peak mem (GB)

Hardware

Chip M3 Ultra (80c)

Memory 512 GB

GPU Cores 80

Software

oMLX v0.2.13

macOS macOS 26.3.1

Context 4,096

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	1,469	105.5	6.7 GB	view
4k	1,491	100.6	7.0 GB	current

Batching Results

Batch Size	TG tok/s	Speedup
1×	105.5	1.00×
2×	181.7	1.72×
4×	246.2	2.33×