Landmark discovery claim · Mathematics
OpenAI releases 722 mathematical manuscripts and selected proof artifacts for community assessment
A public collection of 722 manuscripts in 372 result families, with selected Lean formalizations, reasoning summaries and revision protocols.
Summary
OpenAI published a substantial mathematical claim collection after testing an internal model on approximately 4,000 problems. The repository includes manuscripts, selected Lean artifacts, ten reasoning summaries and checking instructions. It acknowledges uneven verification and possible issues in unformalized results. AGMAI states that its advisory role is neither an endorsement nor an assessment of the results.
AI role
An unreleased model generated research-problem outputs; OpenAI grouped and selected them for release, with exceptions to its standard procedure and some human editing.
Narrative role
Creates a major public assessment workload and an inspectable record of AI-generated mathematical claims. The release is tracked as one collection event, without counting its manuscripts as independently verified discoveries.
Caveat
No Lean compilation or mathematical refereeing was performed in this run. Manuscript counts are not solved-problem counts; related families, assumptions, novelty and overlap with earlier releases require individual assessment.