← Back to community benchmarks

Ornstein-Hermes-3.6-27b-SABER

M3 Ultra (80c) · 512 GB · 8bit · 2026-04-28

Performance

16k

tokens

1,550

PP tok/s

18.8

TG tok/s

10573

TTFT (ms)

32.9

Peak mem (GB)

Hardware

Chip M3 Ultra (80c)

Memory 512 GB

GPU Cores 80

Software

oMLX v0.3.8rc1

macOS macOS 26.3.1

Context 16,384

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	386.6	21.7	29.1 GB	view
4k	421.0	21.3	30.5 GB	view
8k	420.5	20.7	31.2 GB	view
16k	1,550	18.8	32.9 GB	current