← Back to community benchmarks

Qwen3.5-122B-A10B-q4-affine-g64

M5 Max (40c) · 128 GB · 4bit · 2026-03-23

Performance

1k

tokens

918.6

PP tok/s

64.2

TG tok/s

1115

TTFT (ms)

65.6

Peak mem (GB)

Hardware

Chip M5 Max (40c)

Memory 128 GB

GPU Cores 40

Software

oMLX v0.2.20.dev2

macOS macOS 26.3.1

Context 1,024

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	918.6	64.2	65.6 GB	current
4k	1,451	61.4	67.1 GB	view

Batching Results

Batch Size	TG tok/s	Speedup
1×	64.2	1.00×
2×	97.8	1.52×
4×	123.8	1.93×