OpenAI’s unreleased Astra model reportedly solves ten major open math problems
TheZvi · x · 2026-08-04
A post about OpenAI’s unreleased model Astra says it solved ten major open mathematics problems and then formalized the arguments in Lean.
- The author says math is getting easier for LLMs and that Astra appears to be a real step forward.
- OpenAI is reported to have produced new results for ten problems using an internal version of Astra, with human help turning the arguments into manuscripts.
- The model then formalized each argument as a Lean certificate, suggesting a serious jump in theorem-proving capability.
Related event: Rumors of OpenAI's Unreleased Astra Solving Math Problems Spark Debate(6 posts)→
More from Models
- AI went from failing grade-school math to top-level math in just a few years — andrew_n_carr · 2026-08-04
- LeCun debate revisits whether strong code generation is more than pure LLMs — mathemagic1an · 2026-08-04
- Poster says DeepSeek outperformed GPT-5.6 Sol on this output — yacineMTB · 2026-08-04
- Kimi and GLM 5.2 pricing keeps falling as models port across hardware platforms — markjeffrey · 2026-08-04
- OpenAI Reveals How It Built Its Realtime Voice AI System in Just 6 Months — borowcy · 2026-08-04
- Frontier models still fail basic PDE solvers, benchmark post says — GaryMarcus · 2026-08-04