← Back to community benchmarks

Huihui-Qwen3.5-9B-abliterated

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

Performance

1k

tokens

1,120

PP tok/s

106.3

TG tok/s

914

TTFT (ms)

5.8

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 1,024

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	1,120	106.3	5.8 GB	current
4k	1,377	101.8	6.0 GB	view

Batching Results

Batch Size	TG tok/s	Speedup
1×	106.3	1.00×
2×	181.5	1.71×
4×	242.8	2.28×