ValsAI Claims GPT 6 Astra Saturated SRE-Bench
OpenAI researcher Ted Sanders retweeted the claim from @ValsAI about the model result.
TLDR
Ted Sanders, listed as a research engineer at OpenAI, retweeted a post by @ValsAI. The post states that OpenAI’s GPT 6 Astra has effectively saturated SRE-Bench. The benchmark is described in the post as a cybersecurity test of whether models can reverse engineer. The packet contains only this retweet and the quoted claim with no additional posts or corroboration shown.
Combined views
217.1K
6 Sources, first seen ago
