← Back to community benchmarks

Qwen3.5-35B-A3B-UD-Q4_K_XL

M5 Max (40c) · 128 GB · 3bit · 2026-04-04

Performance

16k

tokens

3,878

PP tok/s

92.2

TG tok/s

4224

TTFT (ms)

23.5

Peak mem (GB)

Hardware

Chip M5 Max (40c)

Memory 128 GB

GPU Cores 40

Software

oMLX v0.3.0

macOS macOS 26.4

Context 16,384

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	2,193	108.1	21.5 GB	view
4k	3,624	102.3	22.5 GB	view
8k	3,995	99.4	22.8 GB	view
16k	3,878	92.2	23.5 GB	current
32k	3,321	81.1	24.8 GB	view