OpenAI announced that a cluster of 10,000 autonomous AI agents running on a specialized internal system solved the Navier-Stokes problem—one of the historic Millennium Prize Problems in mathematics—in 88 hours. The proof suggests that fluid dynamics equations can break down under specific conditions as fluid speeds become infinite. Verification of the proof took approximately 17 hours using OpenAI’s GPT-6 Astra model.
However, the announcement prompted controversy after NYU mathematician Tristan Buckmaster questioned whether OpenAI leveraged work he and an Anthropic researcher had stored in OpenAI’s Codex environment. OpenAI denied using the pair’s private data or code, claiming its initiative was sparked by industry rumors and carried out independently.
The achievement highlights the escalating use of massive compute and multi-agent coordination to tackle complex scientific and mathematical challenges. OpenAI stated it does not plan to claim the $1 million Millennium Prize reward, focusing instead on demonstrating the scientific acceleration capabilities of upcoming model architectures.
Why it matters
Multi-agent AI architectures operating over vast compute clusters are rapidly advancing frontier mathematical research.
Data privacy concerns surrounding AI coding assistants remain high when researchers input proprietary or unpublished work.
OpenAI’s milestone underscores a strategic push to position frontier models as transformational tools for scientific discovery.
Source: theguardian.com
