Summary
The collection contains 722 manuscripts organised into 372 related families, with Lean formalisation available for many but not all results. OpenAI warns that some non-formalised material may contain errors and is inviting community scrutiny.
Why it matters
The release provides an unusually broad public test of whether frontier models can produce useful, verifiable mathematical research at scale.
Related coverage
- Mistral opens public preview of trillion-parameter Large 4 model
- Google makes Nano Banana 2.1 generally available
- Danaher plans an AI-powered autonomous laboratory for antibody discovery