← Back to community benchmarks

Huihui-Qwen3.6-35B-A3B-Claude-4.7-Opus-abliterated-vmxfp8

M5 Pro (20c) · 64 GB · 8bit · 2026-07-19

Performance

4k

tokens

2,420

PP tok/s

63.9

TG tok/s

1693

TTFT (ms)

35.1

Peak mem (GB)

Hardware

Chip M5 Pro (20c)

Memory 64 GB

GPU Cores 20

Software

oMLX v0.5.1

macOS macOS 26.5.2

Context 4,096

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
4k	2,420	63.9	35.1 GB	current
8k	2,366	61.2	35.7 GB	view
16k	2,175	59.1	36.6 GB	view
32k	1,827	53.4	38.4 GB	view
64k	1,331	45.8	41.0 GB	view