Unreleased OpenAI model solves 10 major math problems for $2,000 inference cost
Mobile_Distance_9598 · reddit · 2026-08-03
OpenAI's unreleased Astra model has successfully solved 10 long-standing open problems in mathematics and theoretical computer science for an inference cost of roughly $2,000. All proofs are formalized in Lean 4 with machine-checkable certificates and published on GitHub.
Cost and Significance
The results span six mathematical fields, including disproving Connes's rigidity conjecture. Thomas Bloom, a mathematician at the University of Manchester, called the results "big news," rating them more significant than previous breakthroughs.
Controversies and Limitations
- Hidden Costs: The $2,000 covers inference only, excluding training and researcher salaries. However, the author argues that training is a fixed cost, making the marginal cost per attempt approach zero.
- Academic Review: Lean certificates verify formal correctness, not mathematical significance. The results have not been peer-reviewed, and the Astra system itself remains unreleased.
More from Models
- OpenAI's Upcoming 'Astra' Model to Focus on Multi-Agent Collaboration — thesaraharminta · 2026-08-04
- Testing Qwen3.8-Max: Open Models Are Catching Up with Closed Frontier — dair_ai · 2026-08-04
- Kimi K3 Estimated at 2.8T Params, Potentially Distilled from Smaller Opus — gabriberton · 2026-08-04
- Expert Replicates OpenAI Astra Math Results in 24 Hours — Gary Marcus · 2026-08-04
- Claude 3.5 Sonnet and Opus Are Remarkably Terrible at Making Slides — nathanbenaich · 2026-08-04
- Kimi K3 Architecture: Scaling Context, Depth Beyond Bigger MoE — AndLukyane · 2026-08-04