Is Claude Science (Anthropic) Down?

THE MODEL HARDWARE STANDARD -- ANTHROPIC'S SECOND OPEN STANDARD, THIS TIME FOR PHYSICAL LAB EQUIPMENT (2026-08-27, vendor-primary): Anthropic opened a **research preview of the Model Hardware Standard (MHS)**, '**a shared specification for AI agents to safely operate physical devices**', to a first group of scientific research labs and advanced manufacturers. **What it does:** lets agents drive **multiple lab and manufacturing instruments in parallel** -- Anthropic names **microscopes, liquid handlers and robotic arms** -- for work ranging from routine drug-discovery experiments to **laser calibration on a quantum computer**. **The claimed win is integration time: what typically takes a lab weeks or months to set up drops to 'hours or minutes'**, because MHS replaces bespoke per-device integrations with a **standardized driver** exposing simple primitives (`read`, e.g. get temperature; `write`, e.g. set temperature) plus **standard-format discoverability** so devices and agents find each other across networks. The driver also supplies machine characteristics an agent cannot infer from code alone. Anthropic says agents can reason through each step, update parameters in real time, and **in some cases recover from hardware errors without intervention**. **THE STRATEGIC SHAPE IS THE SAME PLAY AS MCP, AND THAT IS THE POINT: MHS is explicitly model-agnostic, works with any device that has a programmable interface, and any agent harness can reach it over standard protocols including the Model Context Protocol.** Anthropic is again defining the interoperability layer rather than a Claude-only feature. **Origin: a collaboration with HHMI Janelia Research Campus.** **STATUS CAVEATS THAT MATTER -- it is an early version, access is application-gated, partners are being recruited specifically to co-develop safety evaluations and best practices for AI operating physical equipment, and it is NOT open source yet ('ahead of making the standard open source').** Nothing here is generally available, and no pricing exists.

LAUNCH (2026-06-30): Claude Science shipped in beta for Claude Pro, Max, Team, and Enterprise users -- an AI workbench pairing a generalist coordinating agent with 60+ curated skills/connectors for genomics, single-cell analysis, proteomics, structural biology, and cheminformatics, plus auditable output history. Anthropic simultaneously announced up to 50 'AI for Science' grants of up to $30,000 in credits (applications through July 15, awards by July 31) and said it will use the product internally for drug-discovery research into rare/neglected diseases. MIT Technology Review called it Anthropic's 'newest flagship product'