For the first time, a large-scale AI model has autonomously accomplished a state-of-the-art theoretical physics computation with minimal human oversight. Anthropic's Claude, armed with just a one-sentence task description and a handful of 'continue' prompts, operated nonstop for several days, incurring a cost of roughly $1,000 to $2,000. It successfully calculated the nine-loop six-gluon scattering amplitude within the framework of planar N=4 super Yang-Mills theory. The outcomes derived from two distinct computational pathways were in harmony, and the model also accurately replicated previously published eight-loop results. Earlier, the Lance team had dedicated years to exploring this domain, calculating the eight-loop amplitude by leveraging S-duality. Upon receiving Anthropic's findings, they proceeded to verify them. Nearly concurrently, the team helmed by He Song from the Institute of Theoretical Physics at the Chinese Academy of Sciences independently deduced the symbol for the nine-loop amplitude using GPT-6 and made publicly available a dataset encompassing amplitudes from two to nine loops. This achievement underscores the dependability and swift enhancement in capabilities of large-scale models when it comes to intricate, cutting-edge physics computations. Presently, these models still depend on established human methodologies and have yet to propose entirely novel physical principles. Researchers in related fields will persist in refining the organization and validation of these results.