← Back to community benchmarks

Kimi-K2.6DQ3_K_M-q8

M3 Ultra (80c) · 512 GB · 4bit · 2026-04-28
Performance
8k
tokens
231.2
PP tok/s
20.1
TG tok/s
35429
TTFT (ms)
443.4
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 222.3 21.9 439.6 GB view
4k 247.7 21.2 441.2 GB view
8k 231.2 20.1 443.4 GB current