← Back to community benchmarks

Qwen3.5-27B-Architect-Polaris-Heretic-mxfp4

M5 Max (40c) · 64 GB · 4bit · 2026-04-18

Performance

16k

tokens

654.1

PP tok/s

29.3

TG tok/s

25047

TTFT (ms)

18.4

Peak mem (GB)

Hardware

Chip M5 Max (40c)

Memory 64 GB

GPU Cores 40

Software

oMLX v0.3.6

macOS macOS 26.4.1

Context 16,384

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	565.5	120.7	18.7 GB	view
4k	573.9	33.6	16.5 GB	view
8k	677.0	32.7	17.1 GB	view
16k	654.1	29.3	18.4 GB	current
32k	583.1	21.8	20.9 GB	view