Unreleased Anthropic model makes progress on Riemann Hypothesis

Anthropic announced that an unreleased model made significant progress on the Riemann Hypothesis, one of mathematics' major unsolved problems standing for over 150 years. The model substantially increased the lower bound of solutions for which the hypothesis holds true. The progress was achieved notably: an Anthropic staff member without significant mathematical training asked the model to 'take a real stab' at proving the hypothesis and then left the model to coordinate the task over a day and a half. The model tested 650 different ideas, coordinating 60 sub-agents and using 31 million output tokens in total. Of the 60 sub-agents, two were responsible for developing the key mathematical ideas. The result was confirmed by two Anthropic in-house mathematicians and formalized using the Lean proof assistant. This finding follows a series of mathematical breakthroughs led by large language models this year.