← Back to community benchmarks

Qwen3.5-4B-Neo

M5 Pro (16c) · 48 GB · 4bit · 2026-04-17

Performance

1k

tokens

1,647

PP tok/s

96.8

TG tok/s

622

TTFT (ms)

3.4

Peak mem (GB)

Hardware

Chip M5 Pro (16c)

Memory 48 GB

GPU Cores 16

Software

oMLX v0.3.6

macOS macOS 26.4.1

Context 1,024

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	1,647	96.8	3.4 GB	current
4k	2,118	91.5	4.1 GB	view

Batching Results

Batch Size	TG tok/s	Speedup
1×	96.8	1.00×
2×	159.0	1.64×
4×	200.2	2.07×