OpenAI Claims Internal Model Solved 100+ Long-Standing Math Problems in a Month
The Decoder · rss · 2026-09-22
OpenAI says an internal model solved more than 100 open math problems after just a month of training, per The Decoder.
- The claim drew criticism from mathematicians over how the results are framed and verified
- In response, OpenAI is backing an independent advisory group at the Institute for Advanced Study
- However, OpenAI excluded its research pace from the group's advisory scope, raising independence concerns
More from Companies & People
- Mistral Reportedly Ditches Frontier Ambitions; Scholar Says No One Outside US and China Is Even Trying — Afinetheorem · 2026-09-22
- Meta rumored to partner with Oracle to bring Meta AI Platform to Oracle Cloud users — testingcatalog · 2026-09-22
- AI researcher recalls being barred from all-male project to avoid 'ruffling feathers' — mmitchell_ai · 2026-09-22
- Mistral's misunderstood play: building Europe's vertically integrated AI infrastructure operator — RemiCadene · 2026-09-22
- Box CEO: AI agents will use software 100X more than humans, reshaping enterprise platforms — _AustinCalvert_ · 2026-09-22
- Remember When Mistral Was the Hottest AI Startup in Europe? — DominiqueCAPaul · 2026-09-22