← Back to community benchmarks

gpt-oss-20b-MXFP4-Q8

M3 Ultra (60c) · 96 GB · 4bit · 2026-04-25
Performance
1k
tokens
1,129
PP tok/s
112.4
TG tok/s
907
TTFT (ms)
11.8
Peak mem (GB)
Hardware
Chip M3 Ultra (60c)
Memory 96 GB
GPU Cores 60
Software
oMLX v0.3.6
macOS macOS 26.3.2
Context 1,024
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,129 112.4 11.8 GB current
4k 1,748 103.5 11.8 GB view
Batching Results
Batch Size TG tok/s Speedup
112.4 1.00×
155.4 1.38×
199.6 1.78×