← Back to community benchmarks

Kimi-K2.5mixed-2.8-bit

M3 Ultra (80c) · 512 GB · 4bit · 2026-03-13

Performance

32k

tokens

158.3

PP tok/s

16.7

TG tok/s

207029

TTFT (ms)

362.6

Peak mem (GB)

Hardware

Chip M3 Ultra (80c)

Memory 512 GB

GPU Cores 80

Software

oMLX v0.2.10

macOS macOS 26.3.1

Context 32,768

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
8k	233.9	24.1	348.1 GB	view
32k	158.3	16.7	362.6 GB	current
64k	104.2	11.9	381.8 GB	view