← Back to community benchmarks

Qwen3.5-4B

M2 Ultra (60c) · 64 GB · 4bit · 2026-03-16

Performance

1k

tokens

1,138

PP tok/s

87.6

TG tok/s

900

TTFT (ms)

4.1

Peak mem (GB)

Hardware

Chip M2 Ultra (60c)

Memory 64 GB

GPU Cores 60

Software

oMLX v0.2.13

macOS macOS 26.3.1

Context 1,024

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	1,138	87.6	4.1 GB	current
4k	1,103	82.7	4.3 GB	view

Batching Results

Batch Size	TG tok/s	Speedup
1×	87.6	1.00×
2×	132.8	1.52×
4×	183.0	2.09×