{"slug": "openai-yang-mills-lean-proof-gross-group-validation-and-public-release", "title": "OpenAI Yang-Mills Lean Proof: Gross Group Validation and Public Release", "summary": "The Yang-Mills Working Group chaired by David Gross published OpenAI's Lean formalization of pure Yang-Mills existence on GitHub as the OpenAI-YM-20261028 repository on November 11, 2026, one day after announcing the release, with an assessment PDF linked from its third committee statement dated November 9. The group's YM repository README states the formalization constructs a Wightman QFT as a scaling limit of Wilson lattice gauge theory, validating existence machinery but not a mass-gap proof, and the committee's first statement said confinement and the mass gap remain open challenges. The Lean package was delivered to the Gross group on October 28, 2026, after OpenAI finished the formalization on October 16 from a natural-language existence proof produced September 9.", "body_md": "On **November 9, 2026**, the **Yang-Mills Working Group** chaired by **[David Gross](https://www.kitp.ucsb.edu/statement-by-the-yangmills-working-group?ref=explainx)** announced it would publish **OpenAI''s Lean formalization** of **pure Yang-Mills existence** the next day. By **November 11**, the **`OpenAI-YM-20261028`** repository on GitHub is **public**, and the group''s **[assessment PDF](https://github.com/yang-mills-group/YM/blob/main/docs/assessment-1.pdf?ref=explainx)** is linked from its **[third committee statement](https://www.kitp.ucsb.edu/sites/default/files/3rd-Statement-About-Quantum-YangMills-Theory_Nov-9.pdf?ref=explainx)**.\n\nThis is the first **machine-checkable release** from the physics shepherding process [OpenAI described on October 26](https://openai.com/index/building-scientific-understanding-with-academic-communities?ref=explainx). It does **not** close the **Clay mass gap** question that [RoundSquare](https://explainx.ai/blog/round-square-lean-yang-mills-mass-gap-peano-openai-what-is-claimed-2026) and social posts keep arguing about.\n\n## TL;DR\n\n| Question | Answer | \n|---|---|\n| What shipped? | Lean code + expert **assessment** on**existence** | \n| Who signed the release? | Douglas, **Gross** , Jaffe, Klebanov, Nekrasov | \n| Mass gap / confinement? | Still **open** , per committee''s prior statements | \n| OpenAI repo | [github.com/yang-mills-group/OpenAI-YM-20261028](https://github.com/yang-mills-group/OpenAI-YM-20261028?ref=explainx) | \n| Read this with | [October triple-result hub](https://explainx.ai/blog/openai-exponential-time-hypothesis-false-kakeya-yang-mills-three-results-lean-collaborations-2026) | \n\n## What the committee actually claims\n\nThe **[November 9 PDF](https://www.kitp.ucsb.edu/sites/default/files/3rd-Statement-About-Quantum-YangMills-Theory_Nov-9.pdf?ref=explainx)** names two artifacts:\n\n1. **Full Lean code** OpenAI submitted**October 28, 2026** — now at**[OpenAI-YM-20261028](https://github.com/yang-mills-group/OpenAI-YM-20261028?ref=explainx)** .\n2. A **validation document** summarizing checks on the October 28 submission. The group's**[YM repository README](https://github.com/yang-mills-group/YM/blob/main/README.md?ref=explainx)** states that, in its opinion, the formalization constructs a**Wightman QFT** as a scaling limit of**Wilson lattice gauge theory** — existence machinery,**not** a mass-gap proof.\n\nThe wording is careful. It validates **existence** under the definitions the committee accepts. It does **not** claim a public proof of **Δ greater than zero** on **R⁴** in the Clay Millennium sense.\n\nThe group''s **[first statement](https://www.kitp.ucsb.edu/sites/default/files/1st-Statement-About-Quantum-YangMills-Theory_Oct-26.pdf?ref=explainx)** already said **confinement and the mass gap remain very important open challenges**. The **[second statement](https://www.kitp.ucsb.edu/sites/default/files/2nd-Statement-About-Quantum-YangMills-Theory_Nov-3.pdf?ref=explainx)** described **definition alignment** work before release.\n\nGreen verification seal evoking machine-checked Lean review of OpenAI Yang-Mills formalization\n\n## Timeline: from model output to GitHub\n\nOpenAI''s **[October 26 post](https://openai.com/index/building-scientific-understanding-with-academic-communities?ref=explainx)** gave these dates for Yang-Mills:\n\n- **September 9, 2026** — natural-language existence proof produced inside OpenAI.\n- **October 16, 2026** — Lean formalization finished.\n- **October 28, 2026** — Lean package delivered to the Gross group (per the November 9 statement).\n\nBetween **November 3** and **November 9**, the committee said it was checking that **definitions match standard statements** and that the **formalization aligns with informal treatment**. The **November 10** release date was announced **one day ahead**, matching the procedure in the second statement.\n\nFor ETH and Kakeya, OpenAI shipped public Lean earlier. Yang-Mills stayed **withheld** until this shepherding finished — the pattern we tracked in **[our launch explainer](https://explainx.ai/blog/openai-exponential-time-hypothesis-false-kakeya-yang-mills-three-results-lean-collaborations-2026)**.\n\n## What is in the GitHub repository\n\nThe published repo description reads **\"OpenAI construction of pure Yang-Mills obtained 10/28/2026\"**. It is **Lean 4** source intended for **community inspection**, not a polished paper.\n\nIf you maintain **scientific agent** pipelines, treat it like any large third-party formal artifact:\n\n- Pin a **Lean toolchain** version before you Mathlib-pin.\n- Read **`#print axioms`** on modules you depend on.\n- Compare **theorem statements** against the informal problem sheet in the assessment PDF.\n\nThe committee credits **Ron Nissim** and **Fred Rajasekaran** on the assessment work. That mirrors how human peer review often splits **authorship** (OpenAI) from **verification reporting** (the group).\n\nChalkboard-style diagram for Yang-Mills existence proof and Clay mass-gap debate context\n\n## Mass gap, RoundSquare, and headline confusion\n\nTwo parallel stories collide in headlines:\n\n| Thread | Claim type | Public artifact | \n|---|---|---|\n| **Gross group release** | **Existence** of pure Yang-Mills in accepted formalization | Lean repo + assessment | \n| **[RoundSquare](https://explainx.ai/blog/round-square-lean-yang-mills-mass-gap-peano-openai-what-is-claimed-2026)** | **Impossibility** of OS axioms + strict gap on**R⁴** | Public Lean + Zenodo preprint | \n| **Clay prize (popular)** | Existence **and** mass gap**Δ greater than 0** | Still contested in public formalizations | \n\nSocial posts that say **\"Peano disproved the $100B proof\"** usually mean RoundSquare''s **infrared argument**, not a refutation of every line in OpenAI''s withheld package. Now that **existence Lean** is public, readers can ask sharper questions: **do the definitions in `OpenAI-YM-20261028` match the assessment''s reading, and do they match the Clay statement Jaffe and Witten published?**\n\nWe are not physicists signing off on QFT measure theory. The **actionable read** for builders is **definition alignment**: automated provers are only as meaningful as the **formal statement** they close.\n\n## How this compares to other lab Lean collaborations\n\nAnthropic and other labs have shipped **formal math** with outside audits — see **[Aaronson–Ambainis conjecture coverage](https://explainx.ai/blog/physical-superintelligence-aaronson-ambainis-conjecture-proof-anthropic-model-lean-claimed-vs-verified-2026)** and **[Opus formalizing CPython](https://explainx.ai/blog/opus-5-5-cpython-lean-formalization-418-bugs-2026)**. The Yang-Mills arc adds **months of withholding** and a **Nobel-laureate committee** as gatekeeper.\n\nThat matters for **agent evals** and **research harnesses**:\n\n- **Time-to-public** can exceed**time-to-generate** by an order of magnitude.\n- **Custody** may sit with**domain committees** , not the lab blog.\n- **Partial releases** (existence first, mass gap later, natural-language write-up later still) are plausible even when Lean exists internally.\n\n## What the assessment says versus what Lean checks\n\nLean does not read like a physics paper. It is a **dependency graph of definitions**, each theorem pinned to axioms Mathlib already exports or to definitions the submission introduces. The committee''s **[assessment PDF](https://github.com/yang-mills-group/YM/blob/main/docs/assessment-1.pdf?ref=explainx)** is the human-readable map between **informal Yang-Mills existence language** and those formal statements.\n\nWhen you audit, copy this checklist:\n\n1. **Locate the main existence theorem** in the repo README or top-level module index.\n2. Run **`lake build`** with the toolchain file the repo pins; do not mix Mathlib versions blindly.\n3. For any claim you care about, run **`#print axioms`** on that theorem — unexpected axioms are a red flag.\n4. Compare the formal **gauge group** ,**measure** , and**Hilbert space** wording against Jaffe–Witten and against the committee PDF, not against Twitter summaries.\n5. File GitHub issues on **definition mismatches** if you find them; that is how public formal math improves.\n\nThe committee thanked **Ron Nissim** and **Fred Rajasekaran** on the assessment — typical for a large formalization where **statement alignment** is as much work as **proof length**.\n\n## AGMAI, withholding, and why release took weeks\n\nThe **[Advisory Group on Mathematics and Artificial Intelligence](https://agmai.org/?ref=explainx)** asked OpenAI in late October to release Yang-Mills outputs **without delay** once Lean existed. The **Gross group** chose a slower path: **validate definitions**, then publish **Lean first** while the **1,147-page natural-language package** waits for another quality gate.\n\nThat split matters for **research agents**:\n\n| Artifact | ETH / Kakeya (October) | Yang-Mills (November) | \n|---|---|---|\n| Lean on GitHub | OpenAI org repos | **yang-mills-group** org | \n| Natural language | Public with Lean | **Still withheld** per committee | \n| Expert gate | Impagliazzo-style CS group | **Physics Nobel-laureate committee** | \n| Prize wording | Complexity / harmonic analysis | **Existence only** in public text | \n\nIf your pipeline ingests **arxiv + GitHub only**, you may now index **existence Lean** but still miss the **full narrative proof** — plan crawlers accordingly.\n\n## Reading list for builders who are not QFT experts\n\nYou do not need to master constructive QFT to extract value:\n\n- **Wilson lattice** intuition — why discretization appears before continuum limits.\n- **Balaban RG program** — the historical line OpenAI says its proof extends.\n- **OS/Wightman axioms** — the package RoundSquare argues is over-constrained on**R⁴** .\n- **Clay problem sheet** — the exact conjunction journalists conflate with**existence alone** .\n\nOur **[Millennium prize AI scorecard](https://explainx.ai/blog/millennium-prize-problems-ai-solved-fact-check-2026)** and **[OpenAI 722-manuscript repo guide](https://explainx.ai/blog/openai-722-math-manuscripts-github-repo-what-to-check-2026)** help separate **headline counts** from **checkable artifacts**.\n\n## What people building with formal methods should do next\n\n**Clone and compile.** Start from the committee''s repo URL, not a fork you cannot trace.\n\n**Read the assessment before the lore.** The PDF is the committee''s **summary of checks**, analogous to a referee report''s \"main theorem verified\" paragraph.\n\n**Separate existence from prize language.** If your product marketing says **\"solved Yang-Mills,\"** you are ahead of the committee''s own **mass-gap** disclaimer.\n\n**Watch for the natural-language 1,147-page package.** The first statement said the group may release **full LLM proofs** once satisfied with quality. That is a **second drop**; today''s release is **Lean-first**.\n\n**Pair with complexity-news hygiene.** The same October batch included **[ETH and Kakeya Lean](https://explainx.ai/blog/openai-exponential-time-hypothesis-false-kakeya-yang-mills-three-results-lean-collaborations-2026)** with different public timelines. Do not assume one release policy fits all three.\n\n## Implications for AI evals and science agents\n\nTeams building ** theorem-proving agents** should log three metrics on this release:\n\n**Time from generation to public Lean** — roughly **seven weeks** from OpenAI''s stated Lean completion to GitHub, dominated by **human committee** time, not compute.\n\n**Custody model** — artifacts live under a **third-party org**, not the lab that generated them. Your agent must follow redirects and mirror policies.\n\n**Partial claim scope** — automated headline tests should flag posts that say **\"Clay solved\"** when sources say **existence only**.\n\nPair with **[AI peer review planted-errors benchmarks](https://explainx.ai/blog/ai-peer-review-planted-errors-benchmark-dawes-institute-2026)** if you score models on **reading referee-style PDFs**, not just compiling Lean.\n\n## Questions readers search after this headline\n\n**Did OpenAI disprove RoundSquare?** Not automatically. RoundSquare targets a **different conjunction** on **R⁴**. Both can be **formally valid** under different statements.\n\n**Can I cite this in production?** Cite the **committee assessment** and **theorem statements you verified**, not uncopied social threads.\n\n**Will API prices change?** No signal in KITP or GitHub releases. This is **science process**, not product pricing.\n\n**Does this help coding agents?** Indirectly — it is another datapoint that **long-horizon formal tasks** can produce **gigabyte-scale artifacts** needing **expert gates** before trust.\n\n## What is still unverified or open\n\n- **Independent mathematicians** beyond the committee may publish critiques of definitions or analytic steps inside the Lean.\n- **Clay Mathematics Institute** has not, in sources we checked, issued a prize decision based on this release.\n- **Mass gap and confinement** remain**open** in the committee''s own framing.\n- **RoundSquare** and**OpenAI existence** may both be**formally consistent** if they formalize**different problem statements** — that is a job for experts, not a headline.\n\n## Related reading\n\n*Repository URLs, committee statements, and OpenAI announcement text are accurate as of November 11, 2026.*", "url": "https://wpnews.pro/news/openai-yang-mills-lean-proof-gross-group-validation-and-public-release", "canonical_source": "https://explainx.ai/blog/openai-yang-mills-gross-group-lean-validation-november-2026?utm_source=rss&utm_medium=feed&utm_campaign=blog", "published_at": "2026-11-11 00:00:00+00:00", "updated_at": "2026-11-11 06:20:33.753537+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-research", "large-language-models"], "entities": ["OpenAI", "Yang-Mills Working Group", "David Gross", "OpenAI-YM-20261028", "Wilson lattice gauge theory", "Wightman QFT", "Clay Millennium Prize", "GitHub"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/openai-yang-mills-lean-proof-gross-group-validation-and-public-release", "markdown": "https://wpnews.pro/news/openai-yang-mills-lean-proof-gross-group-validation-and-public-release.md", "text": "https://wpnews.pro/news/openai-yang-mills-lean-proof-gross-group-validation-and-public-release.txt", "jsonld": "https://wpnews.pro/news/openai-yang-mills-lean-proof-gross-group-validation-and-public-release.jsonld"}}