AI Agents Solve 49 of 50 Physics Problems
A new preprint benchmark, TheoPhysEval, reports that GPT-6-Astra solved 49 of 50 unpublished, research-level theoretical physics problems when web search was available. The study was led by physicists from ICTS, Google DeepMind and other institutions, and covered 11 areas of physics. Without internet access, GPT-6-Astra solved 46 problems. Opus 5.5 also scored 46, with or without web access. The researchers said AI agents’ work helped identify errors in some human-written reference solutions. The realistic-track run cost $116.28. The results suggest AI agents can handle demanding, autonomous scientific problem-solving, but the study is a preprint based on a limited set of 50 questions. It does not test whether AI can generate new research questions. The findings have no direct link to cryptocurrency prices or trading.
Neutral
The study has no direct connection to cryptocurrencies, blockchain networks or digital-asset markets, so it provides no clear trading signal. In the short term, the result could contribute to broader enthusiasm around AI capabilities, a theme that has sometimes supported sentiment toward AI-related crypto projects. However, headlines about AI progress alone have generally not established a reliable, sustained link to token prices; traders usually look for project-specific adoption, revenue, product releases or other measurable catalysts. Over the longer term, wider AI adoption could indirectly affect crypto through demand for computing infrastructure, decentralised AI services or investor interest in related tokens. Those effects remain speculative, and this physics benchmark does not provide evidence of increased blockchain usage or crypto investment. Overall, traders should treat the news as neutral for the market and monitor separate crypto-specific indicators, including trading volume, liquidity, funding rates and project announcements.