PANews, June 20 - Anthropic announced the results of the second phase of "Project Fetch", evaluating the capability improvements of its latest model in real-world robot operations. Conducted in August 2025, the experiment had non-robotics-expert Anthropic employees use off-the-shelf quadruped robots to complete a series of complex tasks, comparing performance between "using Claude model assistance" and "relying solely on humans and the internet". Results showed that under fully autonomous operation of the latest model Claude Opus 4.7, its average speed across all completable tasks significantly surpassed human teams, with execution speed at least 10 times faster.
Anthropic stated that this progress is not the result of specialized robot training, but rather an extension of general large model capabilities, and pointed out that AI is moving from "assisting humans in using tools" to the early stage of "directly operating physical tools", similar to the previous evolution toward agentic programming in software engineering.