Anthropic's Unreleased Claude Model Achieves Breakthrough on Riemann Hypothesis

Anthropic announced that an unreleased research version of Claude autonomously improved the lower bound of the proportion of zeros on the Riemann zeta function's critical line from 41.6% to 67.2%. The model achieved this using Claude Code, generating 31 million tokens and deploying 60 sub-agents over a multi-day run. This marks a significant milestone in AI-driven scientific discovery, demonstrating that LLM-based multi-agent systems can autonomously solve complex, unsolved mathematical problems. The resulting proof was verified by prominent mathematicians and successfully formalized in the Lean theorem prover. The model executed 2,400 shell commands and wrote hundreds of Python scripts, overcoming 650 failed attempts before finding a solution. The mathematical approach involved constructing a function space with a Weil-induced quadratic form and allowing non-diagonal forms to unify positive and negative definite subspaces.