← Back to community benchmarks

Kimi-K2.6DQ3_K_M-q8

M3 Ultra (80c) · 512 GB · 4bit · 2026-05-19

Performance

1k

tokens

220.9

PP tok/s

22.0

TG tok/s

4635

TTFT (ms)

439.6

Peak mem (GB)

Hardware

Chip M3 Ultra (80c)

Memory 512 GB

GPU Cores 80

Software

oMLX v0.3.9.dev2

macOS macOS 26.5

Context 1,024

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	220.9	22.0	439.6 GB	current
4k	246.2	21.5	441.2 GB	view

Batching Results

Batch Size	TG tok/s	Speedup
1×	22.0	1.00×
2×	32.6	1.48×
4×	40.5	1.84×