← Back to community benchmarks

Qwen3.5-122B-A10B-5bit

M3 Max (40c) · 128 GB · 5bit · 2026-04-04

Performance

16k

tokens

1,223

PP tok/s

11.6

TG tok/s

13400

TTFT (ms)

85.2

Peak mem (GB)

Hardware

Chip M3 Max (40c)

Memory 128 GB

GPU Cores 40

Software

oMLX v0.3.2

macOS macOS 26.4

Context 16,384

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
16k	1,223	11.6	85.2 GB	current
32k	1,073	11.3	86.7 GB	view
64k	973.1	11.0	88.5 GB	view