← Back to community benchmarks

inferencerlabs-Qwen3.5-122B-A10B6.5bit

M4 Max (40c) · 128 GB · 6bit · 2026-03-30

Performance

4k

tokens

549.1

PP tok/s

40.9

TG tok/s

7459

TTFT (ms)

95.5

Peak mem (GB)

Hardware

Chip M4 Max (40c)

Memory 128 GB

GPU Cores 40

Software

oMLX v0.3.0rc1

macOS macOS 15.2

Context 4,096

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	466.8	43.3	94.0 GB	view
4k	549.1	40.9	95.5 GB	current

Batching Results

Batch Size	TG tok/s	Speedup
1×	43.3	1.00×
2×	61.2	1.41×
4×	74.0	1.71×