OpenAI Uses 10,000 AI Agents to Produce Proposed Navier-Stokes Solution

OpenAI says a system of roughly 10,000 coordinating artificial intelligence agents has produced a proposed solution to the Navier-Stokes existence and smoothness problem, one of mathematics’ most difficult unresolved questions after nearly 90 years. The effort reached its result in 88 hours and was subsequently formalized and checked using Lean.

According to Coin Bureau, the project consumed enough computational resources to represent roughly $15 million in tokens for an everyday user. OpenAI’s own account gives a larger picture of the operation: the Navier-Stokes effort generated approximately 130 billion output tokens and 2.7 million messages exchanged among agents.

AI Agents Target a Millennium Prize Problem

The Navier-Stokes problem concerns whether smooth three-dimensional fluid flows can remain well behaved indefinitely or instead develop a singularity in finite time. It is one of the seven Millennium Prize Problems established by the Clay Mathematics Institute.

OpenAI said its agents first worked on a simpler mathematical problem before shifting resources toward Navier-Stokes. Different groups pursued separate approaches, with Codex later used to consolidate useful intermediate results. The system reached its reported resolution on Sept. 5, approximately 88 hours after the agents were launched.

The proposed solution describes a vortex that spirals inward and becomes increasingly stretched until the flow develops a finite-time singularity, while the total energy remains finite. OpenAI has released an analytical proof alongside a markets formalization in Lean, enabling the logical steps to be checked computationally.

Formal Verification Does Not End Independent Review

The use of Lean provides an important verification layer, but it does not by itself mean the mathematical claim has received formal recognition. Independent mathematicians still need to examine the proof, its assumptions and whether the result fully financial satisfies the conditions of the Millennium Prize formulation. CoinDesk reported that the proposed solution has already prompted scrutiny over how independently the system arrived at its result.

The scale of the experiment is nevertheless significant for AI-assisted research. Rather than relying on a single model response, OpenAI distributed the mathematical search across thousands of agents, demonstrating a research workflow that treats AI systems more like a coordinated computational workforce.

The next stage will be independent mathematical examination of the proposed proof. OpenAI’s formalized work gives researchers a concrete artifact to inspect, making the coming review crucial to determining whether the claimed breakthrough withstands external scrutiny.

Writer: Victoria Hale
Technology & Blockchain Writer

Victoria Hale writes about blockchain technology, digital infrastructure, and the intersection of emerging technologies with finance. Her articles explore how new protocols and systems are shaping the evolving digital economy.

She prioritises clarity and accuracy when explaining technical developments to a general audience.