OpenAI’s GPT-6.2 Astra hinted Navier-Stokes proof

officechai.com5 min read

An OpenAI perplexity chart published by Chief Scientist Jakub Pachocki suggests that the newly released GPT-6.2 Astra model shares a base model with an internal system that resolved the Navier-Stokes Millennium Prize problem. The chart plots model capabilities against log-perplexity and places GPT-6.2 Astra on the threshold for solving a Millennium Prize math problem. The hint also provides context for the model's strong performance on scientific and technical benchmarks alongside mixed general reviews. Open article →