Astra-Low Tops Eyebench, User Claims
OpenAI researcher retweets post asserting astra-low leads eyebench by wide margin.
TLDR
Ted Sanders, identified in the post as an OpenAI Research Engineer working on LLMs and prompting, retweeted content from @adonis_singh. The visible text reads: "fuck man. astra-low is already beating every single model on eyebench by a wide margin. (including 5.6-sol max) i litera…" The post is presented as a direct quote from its author. No independent confirmation or additional details appear in the packet. The message cuts off mid-word at the end.
Combined views
15
1 Source, first seen 26d ago