OpenAI Uses 10,000 AI Agents to Solve 90-Year-Old Navier-Stokes Math Problem in 88 Hours
OpenAI researchers used 10,000 coordinating AI agents to solve the Navier-Stokes fluid dynamics problem, a mathematical challenge that has stumped scientists for 90 years, in just 88 hours. The achievement, announced in September 2026, marks one of the most significant demonstrations of AI-assisted scientific discovery to date. The result has sparked attribution disputes with independent researchers who say they were working on related approaches.
OpenAI researchers used 10,000 coordinating AI agents to solve the Navier-Stokes fluid dynamics problem in 88 hours, cracking a mathematical challenge that has stumped scientists for 90 years.
The achievement, announced in September 2026, marks one of the most significant demonstrations of AI-assisted scientific discovery to date. The Navier-Stokes equations describe how fluids move and are fundamental to fields ranging from aeronautics to climate modeling. A complete mathematical solution has been one of the seven Millennium Prize Problems, each carrying a $1 million award from the Clay Mathematics Institute.
OpenAI said the system used its new Agents API, which launched in public beta on September 10, 2026. The platform uses a managed Codex harness to handle orchestration, session management, and context compaction, allowing thousands of AI agents to work in parallel on different aspects of a problem.
The result has not been without controversy. Independent researchers have raised attribution disputes, saying they were working on related approaches and that some of the methods used by OpenAI's agents drew on their published work. OpenAI has not publicly responded to those specific claims.
The announcement came the same week that NVIDIA's Nemotron 3 Ultra pipeline achieved a gold-medal score at the 2026 International Mathematical Olympiad, another milestone in AI mathematical reasoning.
OpenAI also released GPT-6 Astra in September 2026, a frontier model capable of complex professional work including cybersecurity and scientific discovery. The company has been pushing AI into semiconductor design as well, successfully completing its first chip design project.
The Navier-Stokes result, if it holds up to peer review, would represent a genuine scientific breakthrough. Mathematicians and physicists are now reviewing the work, and the Clay Mathematics Institute has not yet commented on whether the solution meets the criteria for the Millennium Prize.
For the broader AI industry, the achievement reinforces the case that large-scale multi-agent systems can tackle problems that have resisted human effort for generations.
