OpenAI publishes 722 AI-generated math manuscripts from an unreleased model
OpenAI released 722 mathematical manuscripts in 372 result families to GitHub under Apache-2.0, produced by an unreleased internal frontier model posed ~4,000 open problems, many with Lean proof formalizations.
Why it matters
The largest public corpus of AI-generated mathematical research ever released, with machine-checkable proofs and versioned revision history, hit #3 on Hacker News (945 points, 883 comments) and 7,000+ GitHub stars in a day.
What you could build with it
A research team or edtech startup could build a searchable proof-graph over the 372 families, so mathematicians can find reusable lemmas and formalized Lean proofs instead of re-deriving results. A journal or university could offer an AI-assisted verification service that cross-checks new submissions against the corpus's machine-checked proofs.
Does it hold up?
Too early to judge. The corpus is a day old; mathematicians are still assessing correctness, and only a subset of manuscripts carry machine-checkable Lean proofs, so real-world research utility depends on the verification wave still underway.
Built with OpenAI publishes 722 AI-generated math…
- OpenAI Releases 722 Math Manuscripts From an Unreleased AI ModelUnite.AI · In-depth writeup of the repo structure: 372 families, preprints directory, Lean library, formalization catalogue, and abridged reasoning summaries for 10 results.
- OpenAI announces 722 mathematical discoveries in one goNew Scientist · Coverage with mathematician reactions, including caution from Imperial's Kevin Buzzard that only one of 30 number-theory papers in his field was Lean-verified.
- OpenAI publishes 722 AI mathematics manuscriptsRuntimeWire · Technical breakdown of the release: uneven verification status, versioned correction history, and BibTeX citation blocks per manuscript.
- plby/lean-proofsgithub · 48 ★ · Independent Lean formalizations of Erdos problems, including a disproof built on OpenAI's ten-proofs certificate from the same research program.
Learn more
First spotted on github: source.
More AI models
Mistral Large 4 (Le Chonk): 1.05T-param open-weight MoEMistral's new flagship: a 1.05T-parameter mixture-of-experts multimodal model with 49B active params and 1M-token…models · JEV 0.72ElevenLabs Eleven v4 + v4 Turbo: new emotive TTS architecture with a 100ms real-time variantElevenLabs released Eleven v4 and Eleven v4 Turbo on September 28: a new text-to-speech architecture with inline…models · JEV 0.72VoiceStudio: open-source, fully-local ElevenLabs alternative with voice cloning, dubbing and dictation in 646 languagesVoiceStudio is an open-source, fully-local voice AI studio: voice cloning from a 3-second sample, voice design, video…models · JEV 0.71Mistral Forge: enterprise platform for training and continuously improving proprietary modelsMistral AI launched Forge (Sep 28) — a full-lifecycle model training platform (pre-training, SFT, DPO/ODPO, RL…models · JEV 0.69DeepSeek V4-Flash official API: public beta with upgraded agent capabilities, Responses API and Codex supportDeepSeek's official V4-Flash is now a public beta API with massively upgraded agent capabilities - benchmark scores…models · JEV 0.69kev: open, trainable Jev-like family of small decision models on Qwen3.5/3.8A family of small decision models (0.8B to 27B) built on Qwen3.5/3.8 that reproduces Jev's prefill-only typed-decision…models · JEV 0.68
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.