← Back to community benchmarks

Qwen3.5-27B

M3 Ultra (80c) · 96 GB · 4bit · 2026-03-16

Performance

1k

tokens

421.1

PP tok/s

38.5

TG tok/s

2432

TTFT (ms)

16.6

Peak mem (GB)

Hardware

Chip M3 Ultra (80c)

Memory 96 GB

GPU Cores 80

Software

oMLX v0.2.13

macOS macOS 26.3.1

Context 1,024

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	421.1	38.5	16.6 GB	current
4k	430.1	36.1	17.0 GB	view

Batching Results

Batch Size	TG tok/s	Speedup
1×	38.5	1.00×
2×	69.4	1.80×