Sep. 9 at 10:15 PM
$GOOGL $GOOG OpenAI's agent swarm (10000 Agents) solved the Navier-Stokes problem in 4 days.
4 days lol.
Google DeepMind was working with a team of 20 Mathematicians & AI researchers using the AlphaEvolve and AlphaProof systems to solve the same problem for a whole year, and so were Anthropic researchers.
Plagiarism allegations nonetheless, Google seems to have really fallen badly behind the frontier. It's just one breakthrough after another that OpenAI and Anthropic are sharing using their LLMs.
It used to be that Gemini DeepThink was equally good till December 2025 at IMO/IPhO level problems. However, the 3.5 Pro pre-training run failing really pushed them behind.
Hoping Gemini 4 can bridge this gap. The issue however, remains the inflated price of GOOGL heading into a time of heightened uncertainty.
