GPT-6-Astra solves 49 of 50 physics research problems at $2.40 apiece

A new benchmark shows the best AI agents can solve theoretical physics calculations as accurately as leading specialists in each subfield. The researchers behind it say a missing sense of "taste" is one reason breakthroughs haven't followed yet. Thirty-five researchers from institutions including the International Centre for Theoretical Sciences (ICTS), Harvard University, KU Leuven, and Google DeepMind have released TheoPhysEval.

The benchmark contains 50 unpublished research problems in theoretical physics with verifiable answers. According to the authors, a leading expert in the relevant subfield would need about a day or less to solve a typical problem without AI help. Specialists from neighboring fields would take much longer. The team tested commercially available AI agents and gave them no human hints.

GPT-6-Astra, running in OpenAI's Codex coding environment, came out on top. It found the right answer for 49 of 50 problems, or 98 percent, at an average cost of under $3 per problem and just under $116 for the whole benchmark. The collection deliberately leaves out major open questions and focuses on what the authors call "turn-the-crank" problems, where researchers…