OpenAI announced on August 1, 2026 that an internal version of its upcoming Astra model family, designed to coordinate multiple agents on tasks lasting hours or days, produced solutions to ten previously unsolved problems in mathematics and theoretical computer science, some untouched for over a decade. The results spanned group theory, quantum complexity, lattice cryptography and combinatorics, and were formalized as machine-checkable proofs in the Lean language. Sam Altman has already demonstrated Astra to policymakers in Washington, and OpenAI says the model family will be among the first to go through the US government’s new pre-release AI review process.
