Know a great AI tool or AI news source? Submit it to AiByts
Cline5 min read

11,840 TPS on a single GPU

NVIDIA released Nemotron 3.5 Super VL on October 15, and we reached 11,840 aggregate output tokens/sec on one GB300 with 256 concurrent requests. This blog post is a walkthrough of all the tricks we used to get there. For context, this model is specifically designed to be

By Ara KhanCline
11,840 TPS on a single GPU

Read the full story

Originally published by Cline on Oct 27, 2026.

Read on Cline(opens in a new tab)

Comments

Sign in to join the discussion.

No comments yet — be the first to share your thoughts.