Claude solves amplitude of nine loops in super-Yang-Millls N=4

Claude, from Anthropic, solved a range of nine loops in super-Yang-Millls N=4, a calculation that in practice consumes theoretical physicists for long periods and requires heavy symbolic manipulation. The achievement was described in a case study published by Anthropic and narrated in a post-invitation of physicist and scientific writer Matt von Hippel. The episode repositions the discussion about LLMs: it is no longer about writing code or summarizing texts, but rather about linking mathematical reasoning at the research level.

What's a nine-link amplitude in super-Yang-Millls N=4?

In quantum field theory, the probability of particles interacting is calculated as a series: each term corresponds to Feynman diagrams with an increasing number of loops, and each loop adds an integration over internal momentum. Nine loops means nine chained integrations and an explosive number of diagrams. Super-Yang-Millls N=4 is precisely the preferred laboratory of the theorists: it is a highly symmetric theory, with links to the AdS/CFT correspondence, and therefore serves as a testing field for techniques that later migrate to more realistic theories.

Why is the amplitude calculation of nine loops so difficult?

Each additional loop multiplies combinatorial complexity, not just sum. Symbolic results grow in size and last for pages; experts accumulate techniques for decades and still depend on dedicated software and long machine time. It is not a long account problem: it is a problem of structuring strategy, choosing representations, recognizing symmetry, and checking consistency — exactly the kind of task that separates a calculator from scientific reasoning. See also Claude Opus 4.8: the new artificial intelligence that redefines coding and agentic tasks.

Difference between Claude's achievement and math benchmarks

Olympics tests and sets of university problems measure accuracy in short and self-contained statements. The case of the nine bonds is different: the problem is open, the utterance depends on the accumulated context of the area and the result needs to be checked by those who understand the subject. Anthropic treats the episode as a case study, not as a systematic benchmark — that is, it is evidence of ability, not a rate of correctness measured over hundreds of attempts.

What does Claude's resolution change for Brazilian researchers?

For groups of theoretical and mathematical physics applied in Brazil, pragmatic reading is a reduction in the cost of algebraic work: an assistant who assembles and verifies intermediate passages releases human time for the conceptual part. For those who develop scientific software, the message is that integrating reasoning models to symbolic computing flows ceases to be curiosity and becomes an item of productivity. Anthropic itself has already stated that Claude leads a significant share of the research of new models — 26%, according to the company, in the release reported here.

What Claude's case study still doesn't prove

There is a lack of essential information to judge the robustness of the result: how many attempts were needed, what the computational cost involved, whether the calculation was independently verified by specialists and whether it reproduces in problems of equivalent difficulty. Nor is it clear how much the model recovers from published literature on fishnet integrals rather than derive the solution from zero. They are open questions — and that is precisely why the case should be read as a sign, not as a verdict.

Frequently Asked Questions about Claude and Super-Yang-Mills N=4

What exactly did Claude do?

According to Anthropic and Matt von Hippel's account, the model solved a nine-link amplitude in super-Yang-Mills theory N=4, an advanced calculation of quantum field theory.

Does this mean that AI already replaces theoretical physicists?

Nope. The case demonstrates prolonged mathematical reasoning capacity, but does not eliminate the need for problem formulation, physical interpretation and verification by human experts.

Was the result peer-reviewed?

The material is a case study published by Anthropic, not a peer-reviewed article. Details such as number of attempts, cost and independent verification were not detailed.