<!-- mobian-agent-page publisher="dailydev" canonical="https://daily.dev/posts/astra-helps-physicist-nathaniel-craig-on-a-lattice-definition-of-the-standard-model-4henmjkc4" -->

---
title: Astra helps physicist Nathaniel Craig on a lattice...
description: Physicist Nathaniel Craig worked with OpenAI's GPT-6 Astra to develop a candidate lattice definition of the Standard Model, a problem about 50 years old....
canonical: https://daily.dev/posts/astra-helps-physicist-nathaniel-craig-on-a-lattice-definition-of-the-standard-model-4henmjkc4
twitter:card: summary_large_image
twitter:site: @dailydotdev
og:type: website
og:site_name: daily.dev
og:title: Astra helps physicist Nathaniel Craig on a lattice definition of the Standard Model | daily.dev
og:description: Physicist Nathaniel Craig worked with OpenAI's GPT-6 Astra to develop a candidate lattice definition of the Standard Model, a problem about 50 years old....
og:url: https://daily.dev/posts/astra-helps-physicist-nathaniel-craig-on-a-lattice-definition-of-the-standard-model-4henmjkc4
og:image: https://api.daily.dev/og/posts/4henmJkC4.png
og:image:alt: Astra helps physicist Nathaniel Craig on a lattice definition of the Standard Model
og:image:width: 1200
og:image:height: 630
og:locale: en
---

> ## Documentation Index
> Fetch the complete documentation index at: https://daily.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Astra helps physicist Nathaniel Craig on a lattice definition of the Standard Model

**[Collections](https://daily.dev/sources/collections)** · 2 min read · 4 upvotes · 0 comments

## Summary

Physicist Nathaniel Craig worked with OpenAI's GPT-6 Astra to develop a candidate lattice definition of the Standard Model, a problem about 50 years old. Astra's contribution was the mathematical proof, according to @Hesamation. OpenAI published Craig's account as the first post on its new science blog. Fields Medalist Edward Witten has already published a follow-up paper building on the result, reportedly using Opus 5.5. Witten also sits on the Advisory Group on Mathematics and Artificial Intelligence, which recently pushed OpenAI to publicly release AI math findings for independent review.

## Content

OpenAI says its AI models produced a proof of the existence of pure Yang-Mills theory. South African physicists say that is a long way from solving quantum gravity or delivering a theory of everything.

## What the proof does and doesn't show

A shepherding committee chaired by Nobel laureate David Gross reviewed the work. It found that OpenAI's Lean code proves existence according to standard problem statements. It does not prove a mass gap, and it does not prove uniqueness of the infinite-volume limit. Both are needed for the Clay Mathematics Institute's $1-million Millennium Prize.

Peter Woit at Columbia called the Millennium Prize framing "extremely misleading" in a post titled "The Yang-Mills Publicity Stunt".

## Quantum gravity is a different problem

Kevin Goldstein, a physicist at Wits, calls quantum gravity progress via AI "highly unlikely" soon, because nobody has experimental access to the relevant regime. His Wits colleague Vishnu Jejjala is more open to the idea. He says AI helping with quantum gravity is realistic, but AI solving it is less so, because the target isn't fixed.

## Astra and the Standard Model

On the other side, OpenAI has been showing GPT-6 Astra in theoretical physics. Physicist Nathaniel Craig describes working with Astra to develop a candidate lattice definition of the Standard Model. It is the first post on OpenAI's new science blog, and @gdb shared it as an example of Astra for theoretical physics.

@Hesamation notes that Astra helped with the mathematical proof on a problem about 50 years old. Fields Medalist Edward Witten has since published a follow-up paper building on Craig's work, and he used Opus 5.5 for it.

Witten is also one of the nine members of the Advisory Group on Mathematics and Artificial Intelligence (AGMAI). The group recently pushed OpenAI to fix its AI math proof practices, including a demand to publicly release findings for independent review.

## My read

I like that mathematicians are separating two things: how frontier labs present their AI research, and what AI can actually contribute to science. The Yang-Mills episode looks like the first. Craig's lattice work, with a respected outside physicist building on it, looks more like the second.

## Questions this post answers

### what did OpenAI's GPT-6 Astra contribute to the Nathaniel Craig physics result

Astra contributed the mathematical proof for a candidate lattice definition of the Standard Model of particle physics, a problem that had been open for about 50 years. Physicist Nathaniel Craig worked with the model, and OpenAI published his account as the first post on its new science blog.

_developers evaluating how far AI models reach into real research follow results like this on daily.dev._

### who is the Advisory Group on Mathematics and Artificial Intelligence and what are they demanding from OpenAI

AGMAI is a nine-member group that includes Fields Medalist Edward Witten, and it recently pushed OpenAI to fix how it handles AI math proof claims, including demanding that findings be publicly released for independent review. This is separate from whether AI can actually contribute to science.

_anyone weighing AI research claims against independent scrutiny can track this debate on daily.dev._

### did Edward Witten use OpenAI's model or a different one for his follow-up paper on the Standard Model lattice result

Witten used Opus 5.5, a different lab's model, to build on Nathaniel Craig's result that was developed with OpenAI's GPT-6 Astra, according to @Hesamation. That makes the follow-up a case of one lab's AI output being extended by a researcher using a competing lab's model.

_developers comparing frontier model performance across labs can follow cases like this on daily.dev._

## Community take

How the wider developer community reacted, aggregated from 3 discussions and 545 comments across x (as of 2026-11-17).

**TL;DR:** The announcement itself drew little substantive discussion; most replies are off-topic demands for a new GPT model release, complaints about model naming/availability, or tangential questions, with only scattered genuine engagement praising the lattice result as impressive.

**Sentiment:** 20% positive · 15% mixed · 65% skeptical

**The case for**

- A few commenters called the result genuinely impressive even if the AI's contribution was 'just' the proof.
- Some see it as a meaningful milestone, noting the lattice Standard Model problem has been stuck for decades so even a candidate definition is notable.
- One reply highlighted that human physicists reviewed and vetted the AI-generated proof with domain experts.

**The pushback**

- Many commenters dismissed the post as a distraction from OpenAI's lack of communication about new model releases.
- Several questioned how much of the actual mathematical insight came from the model versus the physicist steering it.
- Some expressed skepticism/frustration that OpenAI publicizes flashy science results while ignoring basic user-facing product complaints.

**By community**

- x (skeptical): Most engagement is frustrated demands for a new model release and gripes about OpenAI's priorities, with only a minority genuinely discussing the physics result.

**Hottest debate:** Whether OpenAI should be investing effort in headline-grabbing science results versus addressing user-facing product and model-release complaints.

**Open questions**

- How much of the lattice construction idea originated from the model itself versus the physicist's steering?
- What dead-end approaches did the model propose before arriving at the one that worked?

**Highlights**

> @OpenAI You guys have time for a science blog, but not to write a single line telling us how the new GPT Astra is progressing. 🫠
> — [TokenGremlin on x · 31 points, 4 comments](https://x.com/TokenGremlin/status/2122128242467340490)

> @OpenAI We meant the other model.
> — [OrangeOrcaDraws on x · 39 points](https://x.com/OrangeOrcaDraws/status/2122155765712715787)

> @OpenAI lattice standard model has been stuck for decades. even a candidate definition is a big deal
> — [LaZeR\_first on x · 1 points](https://x.com/LaZeR_first/status/2122132259826192892)

> @OpenAI Curious how much of the lattice idea came from the model vs Craig steering it. Would love to see the dead ends it proposed too, not just the one that worked.
> — [oxniubi on x · 1 points](https://x.com/oxniubi/status/2122129827876782447)

> @OpenAI Craig had lattice theorists Golterman, Kronfeld and Shamir go through Astra's proof with him.
> — [Cydr4l on x · 1 points](https://x.com/Cydr4l/status/2122128516187926663)

**Source threads**

- [x](https://x.com/Hesamation/status/2122135956014657554) · 1 points · 3 comments
- [x](https://x.com/gdb/status/2122126010691060185) · 0 points · 137 comments
- [x](https://x.com/OpenAI/status/2122123437141553630) · 0 points · 405 comments

## Similar posts on daily.dev

- [Ten advances in mathematics and theoretical computer science](https://daily.dev/posts/ten-advances-in-mathematics-and-theoretical-computer-science-ngbzjedvn) · OpenAI · 1 upvotes · 0 comments
- [AI control is losing the proof race\!](https://daily.dev/posts/ai-control-is-losing-the-proof-race--wkn43rye0) · Medium · 0 upvotes · 0 comments

---

Tags: [#openai](https://daily.dev/tags/openai), [#science](https://daily.dev/tags/science), [#gpt](https://daily.dev/tags/gpt)

[View this post on daily.dev](https://daily.dev/posts/astra-helps-physicist-nathaniel-craig-on-a-lattice-definition-of-the-standard-model-4henmjkc4)

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]}
{"@context":"https://schema.org","@type":"TechArticle","headline":"Astra helps physicist Nathaniel Craig on a lattice definition of the Standard Model","url":"https://daily.dev/posts/astra-helps-physicist-nathaniel-craig-on-a-lattice-definition-of-the-standard-model-4henmjkc4","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/astra-helps-physicist-nathaniel-craig-on-a-lattice-definition-of-the-standard-model-4henmjkc4"},"datePublished":"2026-11-16T01:20:22.851Z","dateModified":"2026-11-17T08:28:49.750Z","description":"Physicist Nathaniel Craig worked with OpenAI's GPT-6 Astra to develop a candidate lattice definition of the Standard Model, a problem about 50 years old....","image":"https://pbs.twimg.com/media/HXNXZXFXYAAdmDi.jpg","thumbnailUrl":"https://pbs.twimg.com/media/HXNXZXFXYAAdmDi.jpg","isAccessibleForFree":true,"articleSection":"Collections","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Collections","logo":"https://media.daily.dev/image/upload/s--fk_6ycEi--/f_auto,q_auto/v1780996001/logos/collections?_a=BAMAMiWQ0","url":"https://daily.dev/sources/collections"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/astra-helps-physicist-nathaniel-craig-on-a-lattice-definition-of-the-standard-model-4henmjkc4","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":4},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"openai,science,gpt","timeRequired":"PT2M"}
{"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Collections","item":"https://daily.dev/sources/collections"},{"@type":"ListItem","position":3,"name":"Astra helps physicist Nathaniel Craig on a lattice definition of the Standard Model"}]}
{"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/astra-helps-physicist-nathaniel-craig-on-a-lattice-definition-of-the-standard-model-4henmjkc4#faq","mainEntity":[{"@type":"Question","name":"what did OpenAI's GPT-6 Astra contribute to the Nathaniel Craig physics result","acceptedAnswer":{"@type":"Answer","text":"Astra contributed the mathematical proof for a candidate lattice definition of the Standard Model of particle physics, a problem that had been open for about 50 years. Physicist Nathaniel Craig worked with the model, and OpenAI published his account as the first post on its new science blog. developers evaluating how far AI models reach into real research follow results like this on daily.dev."}},{"@type":"Question","name":"who is the Advisory Group on Mathematics and Artificial Intelligence and what are they demanding from OpenAI","acceptedAnswer":{"@type":"Answer","text":"AGMAI is a nine-member group that includes Fields Medalist Edward Witten, and it recently pushed OpenAI to fix how it handles AI math proof claims, including demanding that findings be publicly released for independent review. This is separate from whether AI can actually contribute to science. anyone weighing AI research claims against independent scrutiny can track this debate on daily.dev."}},{"@type":"Question","name":"did Edward Witten use OpenAI's model or a different one for his follow-up paper on the Standard Model lattice result","acceptedAnswer":{"@type":"Answer","text":"Witten used Opus 5.5, a different lab's model, to build on Nathaniel Craig's result that was developed with OpenAI's GPT-6 Astra, according to @Hesamation. That makes the follow-up a case of one lab's AI output being extended by a researcher using a competing lab's model. developers comparing frontier model performance across labs can follow cases like this on daily.dev."}}]}
```

