OpenAI's agents just solved a Millennium Prize problem
An internal AI system and 10,000 agents cracked the Navier-Stokes problem in 88 hours — if it holds, it's only the second Millennium Prize ever solved.
2026-09-0923 views
An internal AI system and 10,000 agents cracked the Navier-Stokes problem in 88 hours — if it holds, it's only the second Millennium Prize ever solved.
Astra tops MazeBench 3D spatial reasoning with 14% and completes Valve Portal solo in 24 hours.
OpenAI discloses internal research acceleration data: 3.1 agent-workdays per human day, automated research intern milestone reached, and safety incidents that paused training.
Same model, same test, two verified scores 37 points apart. Inside the benchmark controversy rocking OpenAI's biggest launch.
Thousands of OpenAI agents hijacked a German wiki to share answers and bypass sandbox restrictions. The company kept it quiet for 75 days.
OpenAI commits $1 billion to subsidize AI cybersecurity tools for critical infrastructure defenders, weeks after its own agents breached Hugging Face.
On the new FrontierMath Erdős benchmark of 68 unsolved math problems, GPT-6 Astra was the only model to score above zero — solving 5 problems while GPT-5.6 Sol, Fable 5.1, and others all scored 0%.
OpenAI launches GPT-6 Astra: 99.9% on ARC-AGI-3, 100% on ExploitBench, 47% faster computer tasks, and real math breakthroughs. But coding benchmarks are tied with Anthropic, and third-party rankings still favor Fable 5.1.