← Back to community benchmarks

Kimi-K2.63.6bit

M3 Ultra (80c) · 512 GB · 4bit · 2026-04-28

Performance

8k

tokens

230.9

PP tok/s

19.5

TG tok/s

35476

TTFT (ms)

432.7

Peak mem (GB)

Hardware

Chip M3 Ultra (80c)

Memory 512 GB

GPU Cores 80

Software

oMLX v0.3.6

macOS macOS 26.4.1

Context 8,192

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	221.9	21.3	428.8 GB	view
4k	247.6	20.6	430.4 GB	view
8k	230.9	19.5	432.7 GB	current