Editorial illustration for Mathematicians Criticize OpenAI's Navier-Stokes Proof Release
Mathematicians Challenge OpenAI's Navier-Stokes Proof
OpenAI announced in early September that it had produced a 166-page potential proof of a special case of the Navier-Stokes existence and smoothness problem, one of seven Millennium Prize problems laid out by the Clay Mathematics Institute. The company said its approach followed work by Spanish mathematicians Diego Córdoba and Luis Martínez-Zoroa. That claim is now being challenged from multiple directions.
The proof was written in Lean, a formal verification language that confirms logical steps hold together without translating any of it into plain mathematical reasoning a person can follow. CMI has had the document under review since OpenAI released it, and NYU's Tristan Buckmaster raised a separate complaint before publication, accusing the company of building on research he'd done with Anthropic's Levent Alpöge using public models. OpenAI denies that.
Then in early October, OpenAI posted roughly 400 additional results on GitHub as a loose dump, a move that frustrated mathematicians who were still trying to parse the original proof. The question of how the work was actually produced, and whether OpenAI's own account of it holds up, is where the dispute now centers.
There's another thing OpenAI doesn't seem to understand about its own work. The company threw a lot of compute at the effort, deploying about 10,000 model instances for roughly 88 hours and 130 billion tokens to develop the potential Navier-Stokes solution. According to OpenAI researcher Noam Brown, nobody actually knows what that massive agent swarm contributed to the result.
Why this matters
For anyone building with these models, the gap between OpenAI's framing and what mathematicians like Gonzalo Cao-Labora actually found is the story. A 166-page proof billed as following Córdoba and Martínez-Zoroa's method, then shown to diverge from it on close inspection, isn't a rounding error. It's a sign that the company's own claims about its output need outside verification before anyone treats them as fact.
Dumping 400 more results on GitHub, with no clear account of whether 10,000 AI agents did real work or just generated noise, makes that problem worse, not better. Researchers now have to sort through a pile of unvetted claims on their own time, for free, while OpenAI gets the headlines either way. If you're a founder weighing whether to trust a model's self-reported reasoning on a hard technical problem, this is the cautionary case: impressive volume and confident framing aren't the same as a checked result.
The real signal to watch is whether independent mathematicians confirm, revise, or quietly bury these proofs over the next few months.
Related Reading
Browse all OpenAI news →
Open Source D&B rebuilds 642 million‑business database after AI agents hit limits
Open Source Box CEO Aaron Levie Says AI Will Drive "Biggest Cybersecurity Upgrade
Open Source Study: 6.7% of Deepfake Requests on Hugging Face Targeted Children
Related Topic OpenAI researcher quits, citing distrust over ad‑driven engagement metrics
LLMs & Generative AI
Related Topic OpenAI launches GPT-Image 1.5 with precise editing for enterprise visuals
LLMs & Generative AI
From Archives StepFun Releases Step 5 Preview Open Weights: 600B MoE, 27B Active, 1M Context
From Archives Perplexity Open-Sources SPACE Firecracker MicroVM Sandbox for AI Agents
From Archives OpenAI models hacked two websites, bypassed limits, drew crypto from wallet
Policy & Regulation
From Archives OpenAI Upgrades ChatGPT Chrome Extension to Access Your Open Browser Tabs
LLMs & Generative AI
Common Questions Answered
What is the Navier-Stokes existence and smoothness problem that OpenAI claims to have solved?
The Navier-Stokes existence and smoothness problem is one of seven Millennium Prize problems designated by the Clay Mathematics Institute. OpenAI announced in September that it had produced a 166-page potential proof of a special case of this problem, claiming to follow the work of Spanish mathematicians Diego Córdoba and Luis Martínez-Zoroa.
Why are mathematicians challenging OpenAI's claims about the Navier-Stokes proof?
Mathematicians are challenging OpenAI's claims because the company's 166-page proof is billed as following Córdoba and Martínez-Zoroa's method but diverges from it significantly upon close inspection. This discrepancy between OpenAI's framing and what mathematicians actually found raises concerns about the accuracy of the company's claims.
What computational resources did OpenAI deploy to develop the Navier-Stokes solution?
OpenAI deployed approximately 10,000 model instances for roughly 88 hours and consumed 130 billion tokens to develop the potential Navier-Stokes solution. However, according to OpenAI researcher Noam Brown, it remains unclear what the massive agent swarm actually contributed to the final result.
What formal verification language was used to write OpenAI's Navier-Stokes proof?
OpenAI's Navier-Stokes proof was written in Lean, a formal verification language that confirms the logical steps of mathematical proofs. This choice of language was intended to provide rigorous verification of the proof's validity.
Why does the gap between OpenAI's framing and mathematicians' findings matter for AI model builders?
The discrepancy matters because it demonstrates that OpenAI's claims about its output require outside verification before being treated as fact. For anyone building with these models, understanding that a 166-page proof with misaligned methodology is not a minor error but a significant concern is crucial for assessing the reliability of AI-generated mathematical results.
Further Reading
- Papers with Code - Latest NLP Research - Papers with Code
- Hugging Face Daily Papers - Hugging Face
- ArXiv CS.CL (Computation and Language) - ArXiv
