Otto365 AI Brief #010

OpenAI: new maths that a computer can check

How this was made: Otto365 is an animated character and his voice is generated by AI. The video tells the same story as the text on this page; the transcript lists every spoken word and the main lines shown on screen. Spotted an error?
  • Published
  • By the Otto365 AI Brief. Checked and approved by Peter Ferguson, Editor, on . The script was drafted with AI assistance, then every sentence was checked against the sources below.
  • Announced 6 Oct 2026
  • Voiced by AI

What happened

OpenAI's internal model has produced new maths, with proofs a computer can check. OpenAI announced a broad range of new mathematical results from an internal model on the 6th of October 2026. They sit on GitHub, and OpenAI says the model itself is still internal.1

Three things to know

  1. Scale: 722 manuscripts First, the scale. That is 722 manuscripts across 372 result families, reported by Tech Insider, with roughly 4,000 problems tested. OpenAI says the average result took roughly three hours of ChatGPT Pro thinking.12
  2. Lean-checked proofs Second, checking the work. Many proofs are formalised in Lean, a programming language that lets a computer check a proof. OpenAI also publishes 10 summaries of the model's reasoning and statistics on attempted problems.1
  3. Review still underway Third, the scrutiny. OpenAI took advice from an independent advisory group at the Institute for Advanced Study. Outside mathematicians are not convinced the claims hold up, reported by Tech Insider, and community verification is still underway.12

Why it matters to you

Otto365's take (opinion)

So, my take. The reasoning in everyday AI tools keeps improving, and output checked by software is where this is heading. For a UK small business, the lesson is: verify what a chatbot tells you.12

Do this week

Try this: verify one AI answer against its original source.

Sources

  1. OpenAI: Sharing AI progress in mathematics (opens in a new tab) Primary source, accessed 7 Oct 2026
  2. Tech Insider: Tao Group Scrutinizes OpenAI's 722 Math Claims [2026] (opens in a new tab) News report, accessed 7 Oct 2026
The quotes we checked (13)

  • We’re releasing a broad range of new mathematical results produced by an internal frontier model.

    OpenAI · source 1

  • October 6, 2026 Sharing AI progress in mathematics

    OpenAI · source 1

  • we’re publishing the results in a GitHub repository, with protocols for paper revisions and citations.

    OpenAI · source 1

  • we are sharing formalizations of many of the proofs in Lean, a programming language that allows mathematical proofs to be checked by a computer.

    OpenAI · source 1

  • We will update the repository with more formalizations as we obtain them.

    OpenAI · source 1

  • These include 10 summaries of the model’s reasoning, estimations of compute spent in terms of Pro usage on ChatGPT, and statistics about the number of attempted problems.

    OpenAI · source 1

  • The average result used the equivalent compute of roughly three hours of ChatGPT Pro thinking.

    OpenAI · source 1

  • we’ve been consulting with the independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study to develop best practices, and we have drawn on their advice and public recommendations to inform how we release these results.

    OpenAI · source 1

  • are working to responsibly release the model that produced these results.

    OpenAI · source 1

  • 722 manuscripts, organized into 372 result families, reported by GIGAZINE and The Indian Express as the headline figures.

    Tech Insider · source 2

  • OpenAI tested its internal frontier model against roughly 4,000 mathematical problems, with the average published result requiring computing power roughly equivalent to about three hours of ChatGPT Pro reasoning.

    Tech Insider · source 2

  • Mathematicians outside the company are not convinced the claims hold up to the standard of proof the field has used for centuries

    Tech Insider · source 2

  • The manuscripts are public before the community-wide verification process has run its course.

    Tech Insider · source 2

Transcript

Full transcript

OpenAI's internal model has produced new maths, with proofs a computer can check.

Play from 0:05 OpenAI announced a broad range of new mathematical results from an internal model on the 6th of October 2026. They sit on GitHub, and OpenAI says the model itself is still internal.

Play from 0:18 First, the scale. That is 722 manuscripts across 372 result families, reported by Tech Insider, with roughly 4,000 problems tested. OpenAI says the average result took roughly three hours of ChatGPT Pro thinking.

Play from 0:34 Second, checking the work. Many proofs are formalised in Lean, a programming language that lets a computer check a proof. OpenAI also publishes 10 summaries of the model's reasoning and statistics on attempted problems.

Play from 0:49 Third, the scrutiny. OpenAI took advice from an independent advisory group at the Institute for Advanced Study. Outside mathematicians are not convinced the claims hold up, reported by Tech Insider, and community verification is still underway.

Play from 1:04 So, my take. The reasoning in everyday AI tools keeps improving, and output checked by software is where this is heading. For a UK small business, the lesson is: verify what a chatbot tells you.

Play from 1:18 Try this: verify one AI answer against its original source.

I'm Otto365.

Follow the Brief.

On screen

Play from 0:00
Otto365 AI Brief #010, OpenAI, 6 Oct 2026
OpenAI's model did maths a computer can check
Problem
Internal model
Lean check
GitHub repo
Play from 0:05
OpenAI's maths, checked
Announced 6 Oct 2026
What is on GitHub
New maths results
Lean proofs
10 reasoning summaries
Compute estimates
Source: OpenAI, 6 Oct 2026
Play from 0:18
Scale: 722 manuscripts
Tested: 4,000 problems
Published: 722 manuscripts, 372 families
Roughly three hours
ChatGPT Pro thinking
Per average result
Play from 0:34
Lean-checked proofs
Model's proof
Written in Lean
Computer check
GitHub repo
Also in the repository
10 reasoning summaries
Compute estimates
Attempted problem stats
More Lean proofs as obtained
Play from 0:49
Review still underway
OpenAI says: New results, Lean checks many
Mathematicians: Not yet convinced, Review underway
Institute for Advanced Study
Advice on sharing the results
Independent of OpenAI
Play from 1:04
Verify chatbot answers
AI reasoning
Software check
Expert review
Then rely on it
Is this maths result proven?
Many proofs: checked in Lean by a computer
Outside mathematicians: not convinced yet
OpenAI
Tech Insider
Play from 1:18
Check one AI answer at source
Voiced by AI. Checked by humans. Apps 365 sells AI and Microsoft 365 services. Not endorsed by companies named.

Voiced by AI. Checked by humans. Apps 365 sells AI and Microsoft 365 services. Not endorsed by companies named. Report an error.