The AI model Claude Opus 4.7 handled robot dog control 20 times faster than humans — a new leap in autonomy
The artificial intelligence market continues to demonstrate explosive growth in capabilities, and a fresh experiment within the Project Fetch initiative is clear proof of this. Anthropic's Claude Opus 4.7 model not only handled tasks of configuring and controlling a four-legged robot—it did so 20 times faster than teams of human engineers.
In August 2024, when the project first launched, company employees without specialized robotics experience attempted to program a robot dog using AI. At that time, the neural network served only as an auxiliary tool. Now, in the second testing phase, Claude Opus 4.7 worked almost autonomously, under minimal researcher supervision.
What did the model do independently?
The neural network completed a full cycle of tasks that previously required the involvement of an entire team:
- Connected to video sensors and LiDAR;
- Wrote a program for manual robot control;
- Created a motion trajectory monitoring system;
- Configured an object recognition algorithm.
The performance metrics are impressive: Opus 4.7 proved to be 18 times faster than a team using previous AI versions, and 37 times faster than humans working without chatbot assistance. At the same time, code quality also improved—its volume turned out to be 10 times smaller than that of human teams. This speaks not only to speed but also to the efficiency of solutions.
Notably, Anthropic did not implement specialized algorithms for controlling the hardware. According to the developers, the progress in robotics became a side effect of the general scaling of language models. This is an important signal for the industry: universal AI systems are beginning to master the physical world without narrow tuning.
However, there were limitations. Despite success in navigation, Claude encountered difficulties in performing precise physical actions. The model managed to guide the robot to a target but failed at the task of gently pushing a ball—this requires complex real-time feedback, where humans still hold an advantage.
At Anthropic, they believe the industry is entering an era of "physical AI agents." In the near future, neural networks will be able to control standard equipment as naturally as they now write code.
Expert opinion: The breakthrough of Claude Opus 4.7 is not just a test but a marker of a paradigm shift. When AI begins to independently program robots and do so an order of magnitude faster than humans, we move from "tool" to "agent." However, the problem of fine motor skills reminds us that full replacement of humans in physical labor is still far off. It is the combination of AI's cognitive power and human sensorimotor intelligence that will become the next battlefield.