Posted on arXiv on Oct 12, 2026 (source: arxiv.org).
Google DeepMind's AlphaProof Nexus uses LLM agents to search for formal proofs in Lean, with the Lean compiler checking every step. The peer-reviewed paper, published in Science on October 8 after a May arXiv preprint, reports autonomously solving 9 of 353 attempted Erdos problems, including two open for 56 years, and proving 44 of 492 open OEIS conjectures. DeepMind says the full agent cost a few hundred dollars of inference per problem and also helped on problems in algebraic geometry, optimization, quantum optics and graph theory.
Items older than 72 hours never go in Top stories and show their original date.