tweetindex

AI Security Institute (AISI) @AISecurityInst · 27 Aug 2026

Some frontier AI evaluations require hundreds of millions of tokens, and every wasted trial has a real cost. We’re introducing optstop: our open-source tool that stops an evaluation where estimates are already precise, and keeps running where they're not. 🧵 https://t.co/kKHWjPBmED
22 651Views
228Likes
25Reposts
7Replies
7Quotes
135Bookmarks

Is that a lot?

0.22×vs this author's median102 715 views is typical
20Percentile for this authorof 5 recent posts
7.82×vs 10K–100K median2 896 views is typical
111.19%Reachviews ÷ followers
1.18%Engagement rateof viewers reacted

Compare with the benchmark table →

Open on X →