← Back to community benchmarks

gpt-oss-20b

M3 Ultra (80c) · 512 GB · unknown · 2026-06-13

Performance

4k

tokens

2,193

PP tok/s

122.6

TG tok/s

1868

TTFT (ms)

13.3

Peak mem (GB)

Hardware

Chip M3 Ultra (80c)

Memory 512 GB

GPU Cores 80

Software

oMLX v0.4.4rc1

macOS macOS 26.5.1

Context 4,096

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	1,078	128.2	13.3 GB	view
4k	2,193	122.6	13.3 GB	current

Batching Results

Batch Size	TG tok/s	Speedup
1×	128.2	1.00×
2×	118.5	0.92×
4×	205.7	1.60×