Mathematicians demand real proofs from leading AI labs — OpenAI publishes 722 articles

Mathematicians are calling on leading AI labs to publish verifiable proofs instead of blog posts. OpenAI responded with a repository containing 722 articles grouped into 372 results.
Mathematicians called on leading AI labs to publish real proofs that can be checked, rather than blog posts.
OpenAI responded with the openai/math repository, which contains 722 articles grouped into 372 results.
According to the company, the materials were prepared by a new internal model that has not yet been released. Standard mathematical tests proved too easy for it, so the model was given about 4,000 unsolved research problems. Some of them had resisted mathematicians for decades.
On average, each result required about three hours of computation. The model formatted its answers as scientific articles, and some of them claim to solve very well-known problems.
Why it matters
AnalysisThis initiative shows how AI can generate formalized mathematical artifacts, but without open and verifiable proofs they remain only claims. For researchers, it is important that AI labs publish materials that can actually be checked.
Discuss in community
Share your questions and insights with developers
Related in Ecosystem
Practical guides, core terminology, and agent skills
OpenAI Releases GPT 6.2 Astra and Sol
OpenAI has launched GPT 6.2 Astra and Sol, with Astra topping agent benchmarks and offering more competitive token pricing.
GPT-6.2 Astra and Sol: pricing, benchmarks, and mixed community reaction
OpenAI’s GPT-6.2 Astra and Sol bring cheaper Astra pricing and improved Sol quality, but early community feedback points to loops, readability issues, and weaker jailbreak resistance.
Part II. AI Agent Architecture: From Model to Memory and Tools
A deep engineering dive: tokenization, KV-cache, sampling determinism, the training stack, quantization, context engineering, the safety stack, production RAG, ReAct, and multi-agent swarm systems.
OpenAI GPT (Flagship Models of the GPT Series)
The primary universal line of large language models from OpenAI (GPT-4, GPT-4o). Optimized for complex text analysis, programming, creativity, and daily intellectual tasks.