Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Max token throughput on 64K/64K serving scenario 8xH100 node 1.0

9.3Max Throughput (K tok/s)

gpt-oss-puzzle-88B

3.7885.2196.658.081Feb 12, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.02
9.3
2026.02
6.9
2026.02
6.5
2026.02
5.8
2026.02
4