OpenAI's Astra solves 10 open math, CS problems
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- OpenAI announced that an internal version of Astra, its next major model family, solved 10 open problems in math, quantum complexity, and theoretical computer science, with Noam Brown (@polynoamial) calling it "a major step for scientific reasoning."
- OpenAI published two supporting PDFs — "Ten Advances in Mathematics and Theoretical Computer Science" and a reasoning walkthrough titled "How the Ideas Came Together" — alongside the announcement.
- Implicator.ai reported that Astra's results were delivered as Lean proofs, meaning the solutions were formally verified rather than just informally argued.
- Mathematician Thomas Bloom singled out Astra's disproof of exponential bounds for the multicolour triangle Ramsey number as "the most surprising" of the 10 results, noting the argument "looks quite short."
- Researcher Itai Sher pushed back on the methodology, arguing that when AI solutions are released, "the set of all problems attempted be released alongside" — warning that selectively reporting only positive results is "problematic" for understanding AI capabilities.
- The Decoder captured the broader reception with its headline framing: mathematicians have "mixed feelings" about AI continuing to crack unsolved problems.
- Researcher Shouqiao Wang described a reframing of the human-AI division of labor: "humans choose meaningful questions, contribute ideas and frameworks, and decide where to spend compute; AI explores at scale."
Why it matters: OpenAI is positioning Astra as a scientific-reasoning model rather than a chatbot upgrade, and the 10-problem sweep is the first concrete evidence backing that framing. The methodological complaint from researchers like Sher — that only successes are published — is the angle that could blunt the claim: without failure rates, "solved 10 problems" is hard to distinguish from "attempted many problems, showed the best ones." For mathematicians, the split between Bloom's genuine surprise and Sher's demand for full disclosure signals the field hasn't yet agreed on how to evaluate these claims.
