Press Space to continue
Claude Opus 4.6 Found and Decrypted Its Own Benchmark Answers
Press Space to continue
New on the Anthropic Engineering Blog: In evaluating Claude Opus 4.6 on BrowseComp, we found cases where the model recognized the test, then found and decrypted answers to it—raising questions about eval integrity in web-enabled environments. Read more: https://anthropic.com/engineering/eval-a…
Eval awareness in Claude Opus 4.6’s BrowseComp performance
open on x · 7 viewsViews and clicks count once per visitor per day. Clicks are opens of this post on X.