Anthropic previews Model Hardware Standard
Anthropic has opened a gated research preview of the Model Hardware Standard, a specification that lets AI agents discover and operate lab and factory instruments, cutting integration work from weeks to hours.
Anthropic said it is opening a research preview of the Model Hardware Standard (MHS), a shared specification for AI agents to operate physical devices, to an initial group of scientific labs and advanced manufacturers. The work began as a collaboration between Anthropic and HHMI Janelia Research Campus.
MHS targets a specific bottleneck: instruments such as microscopes, liquid handlers and robotic arms each expose their own programming interface, and connecting them typically takes labs weeks or months of bespoke integration work. Anthropic says MHS reduces that to hours or minutes.
The standard centers on a driver built around simple primitives — "read" (get temperature) and "write" (set temperature) — plus a discovery format so devices and agents can find each other across networks. Natural-language tags let users record details not visible in code, such as a robot arm's weight, which the driver compiles into a reference file covering what a device measures, what is adjustable, and what safety limits apply. Agents control hardware three ways: MCP, a command line interface, and code files. MHS works with any device that has a programmable interface and is model-agnostic.
Anthropic describes Claude aligning a laser by adjusting it, observing the result through a camera, then packaging the routine into a deterministic script.
Access is by application. Anthropic says it is working with partners on safety evaluations and best practices before open-sourcing the standard; it has not given a timeline for that release, and partner results reported so far are self-reported.