← Back to community benchmarks

GLM-5.2

M3 Ultra (80c) · 512 GB · 4bit · 2026-07-23

Performance

1k

tokens

183.1

PP tok/s

19.1

TG tok/s

5593

TTFT (ms)

390.9

Peak mem (GB)

Hardware

Chip M3 Ultra (80c)

Memory 512 GB

GPU Cores 80

Software

oMLX v0.4.5.dev1

macOS macOS 26.4.1

Context 1,024

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	183.1	19.1	390.9 GB	current
4k	206.6	17.9	392.8 GB	view
8k	179.5	17.8	394.4 GB	view
16k	100.9	17.6	395.3 GB	view
32k	90.0	15.3	395.9 GB	view
64k	85.8	15.0	399.4 GB	view