The AI agent Claude Opus 4.7 outperformed humans by dozens of times in controlling a robot dog.
We are witnessing another breakthrough at the intersection of artificial intelligence and robotics. As part of the updated Project Fetch experiment, the Claude Opus 4.7 model demonstrated the ability to autonomously program and control a four-legged robot, completing tasks 20 times faster than a team of human engineers.
For context: back in August 2024, Anthropic employees with no robotics experience used AI only to accelerate solution finding. Now, the situation has changed dramatically. In the new testing phase, Claude Opus 4.7 operated almost entirely autonomously, under minimal researcher supervision. The neural network independently performed a full cycle of operations:
- connected to video sensors and LiDAR;
- wrote a program for manual control;
- created a trajectory monitoring system;
- configured an object recognition algorithm.
The numbers are impressive: Opus 4.7's performance was 18 times higher than that of a team using previous AI versions, and 37 times higher than humans working without chatbot assistance. Moreover, the code generated by the neural network turned out to be 10 times more compact than what engineers wrote. This speaks not only to speed but also to the efficiency of the solution architecture that AI offers.
The key conclusion drawn by developers: progress in robotics has essentially become a side effect of the general scaling of language models. Anthropic did not introduce any specialized algorithms for controlling hardware. The model simply learned to understand and perform physical tasks through overall cognitive improvement.
However, there were limitations. Claude still faces serious difficulties with precise physical manipulations. During the experiment, the robot successfully reached its target but could not gently push a ball to the right spot. Such actions require complex real-time feedback — an area where humans still maintain superiority.
At Anthropic, they believe the industry is entering an era of "physical AI agents." In the near future, neural networks will be able to work with standard tools and equipment as effectively as they currently work with software code. This opens up enormous prospects for automation in manufacturing, logistics, and service robotics.
I should add that against the backdrop of these achievements, on June 13, Anthropic was forced to suspend access to the Fable 5 and Mythos 5 models due to a directive from the U.S. government under export controls. This further underscores how strategically significant AI technology is becoming on a global level.
Expert opinion: The breakthrough of Claude Opus 4.7 in robot control is not just another benchmark. It is a signal that the boundary between the software and physical worlds is blurring faster than many expect. Investors and developers should prepare for the fact that within 2-3 years, we will see the first commercial implementations of fully autonomous AI agents in industry.