OpenAI says its internal model solved over 100 long-standing math problems after just a month of training
100+ unsolved math problems. One month of training. OpenAI's internal model just reset what we thought was possible in automated theorem-solving.
Why it matters
OpenAI demonstrated a significant capability leap in mathematical reasoning—a frontier benchmark for reasoning models—but the company is managing external scrutiny by funding (but limiting) independent review at the Institute for Advanced Study.
The key facts
6 to knowOpenAI internal model solved 100+ long-standing open math problems
Training duration: approximately one month
Company facing criticism from mathematicians on claims/methodology
OpenAI backing independent advisory group at Institute for Advanced Study
Company has excluded its research pace from advisory group's scope
Suggests reasoning capability breakthrough, but governance/transparency questions remain
Go to the source
The Decoderthe-decoder.com
Publisher excerpt: OpenAI says a new internal model solved more than 100 open math problems after just a month of training. Facing criticism from mathematicians, the company is backing an independent advisory group at the Institute for Advanced Study. But OpenAI has excluded its research pace from the group's…