MCP, or Model Context Protocol, is an emerging open standard for AI model interfacing and orchestration. It enables AI agents and models to communicate contextually and operate with shared understanding, allowing for more dynamic, composable, and powerful AI applications.

The Block MCP is a platform that provides the most comprehensive and professional collection of MCP servers. Users can access these servers via the cloud, paying based on token usage.

Pricing is based on token consumption. Each MCP server has a predefined cost per token, and users will be charged accordingly.

MCP servers serve as modular, context-aware service endpoints that AI agents can call to perform specialized tasks. By enabling context sharing and reusable workflows across models and systems, MCP servers unlock a new layer of composability, interoperability, and decentralization in AI infrastructure.

You can directly call MCP servers from the cloud and use them via our platform client, ensuring a seamless and efficient experience. Alternatively, you can also deploy the servers locally and use them.