← Back to community benchmarks

Huihui-Qwen3.5-27B-Claude-4.6-Opus-abliterated

M3 Ultra (80c) · 512 GB · bf16 · 2026-04-06

Performance

8k

tokens

432.1

PP tok/s

12.1

TG tok/s

18958

TTFT (ms)

53.8

Peak mem (GB)

Hardware

Chip M3 Ultra (80c)

Memory 512 GB

GPU Cores 80

Software

oMLX v0.3.4

macOS macOS 26.4

Context 8,192

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	378.5	12.5	51.7 GB	view
4k	427.7	12.4	53.2 GB	view
8k	432.1	12.1	53.8 GB	current

Batching Results

Batch Size	TG tok/s	Speedup
1×	12.5	1.00×
2×	13.8	1.10×
4×	27.3	2.18×