Peaked at #3 Now #3
GPT-6 Astra Tops Physics Benchmark, Experts Corrected
Freddy Vega says GPT-6 Astra initially scored +80% on a physics benchmark, then experts confirmed it was right and the new score became 98%.
Key points
- GPT-6 Astra scored +80% on a Physics benchmark and insisted the expert results were wrong and its calculations were right.
- A team checked and found GPT-6 Astra was correct; the new score is 98%.
- @Kaju_Nut said the initial evaluation put GPT-6-Astra in the late eighties, followed closely by Opus 5.5, and that experts later found subtle errors in their own calculations.
- @kimmonismus said TheoPhysEval's authors found errors in the reference solutions and confirmed Astra was correct, and that using Codex with tools and web access it scored 49/50 (98%).
Key points and the reaction summary are written by AI from the posts on this page. Check the original post. How we use AI
Original post
What others are saying
2 moreTop replies on X
Reaction on X: Replies largely welcome the result, one framing it as a shift from "shut up and calculate" to "slow down and think," and another joking that Astra graded the graders. One reply questions how models this good still fail to follow instructions in coding harnesses. The original poster did not respond.
Sign in to see 4 top replies from X
From @Sayan_Quantum, @tchlux, @Kaiser1710 and others. Spam removed, with English and Chinese translations.
Discussion 0
Sign up Sign in to join the discussion
No comments yet. Start the conversation.