According to CoinMeta, Anthropic has launched a model hardware standard ( MHS ) in the form of a research preview, enabling AI and agent to directly operate real-world machines, including microscopes, liquid processors, robotic arms, and laser systems in quantum computers. In the past, connecting devices from different manufacturers usually took weeks or even months, but MHS can compress this process to just a few hours or even minutes. As a unified hardware interface, devices, once connected, will inform agent of their own functions, status, and security limitations in a standardized format. Actual tests have shown that the experimental automation system set up by Carnegie Mellon University in about 8 hours became approximately three times faster. Quantum computing company Quera enabled Claude to automatically optimize laser recovery programs with a success rate of 99.3%. Currently, MHS has not made the standard public yet; it is only available for application by some research institutions and manufacturers. Anthropic plans to open-source it after completing more security tests.