Anthropic’s latest AI model has reportedly reached the top of the Super-Agent benchmark, a grueling test of whether an AI system can take a real-world code repository and run it from scratch without ...