Models Plus

You can access comprehensive AI model metadata through an MCP (Model Context Protocol) server that exposes both an HTTP API and native MCP integration. This enables your assistants to search, filter, and retrieve detailed model and provider information in real time, enabling natural-language queries and seamless model selection.

How to use

Connect your MCP-enabled assistant to the server at https://modelsplus.quivr.tech/mcp. You can use either the HTTP endpoint or an MCP client to interact with the server, depending on your setup. With an MCP client, you’ll invoke tools to search for models, retrieve model details, and inspect providers. For example, you can configure your MCP client to point to the MCP URL and then issue high-level queries like finding models by capability, cost, or provider.

If you are using Claude Desktop, add the MCP server configuration to your claude_desktop_config.json as shown here.

{
  "mcpServers": {
    "models-plus": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/sdk", "server", "https://modelsplus.quivr.tech/mcp"]
    }
  }
}

How to install

Prerequisites you need before starting: Bun runtime (version 1.2.21) as the runtime and package manager, and Node.js types (bundled via SST) for tooling compatibility.

Step-by-step commands to set up and run locally.

# Install dependencies
bun install

# Generate JSON assets from vendor data (API side)
cd packages/api && bun run generate

# Build for production
bun run build

Additional MCP notes

MCP integration provides native support with tools to search models, get model details, and query providers. You can use the CLI-based setup for an MCP client and point it at the public MCP URL for modelsplus.

Example MCP client setup for Claude Desktop shows how to register the public MCP server.

{
  "mcpServers": {
    "models-plus": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/sdk", "server", "https://modelsplus.quivr.tech/mcp"]
    }
  }
}

Notes and tips

No authentication is required to access the public MCP API. You can explore capabilities at the MCP URL and use the listed tools to perform model searches, detailed lookups, and provider inquiries.

Available tools

search_models

Advanced search and filtering for AI models across the catalog, enabling queries by capabilities, cost, context length, modalities, and more.

get_model

Retrieve detailed information about a specific model, including capabilities, costs, and limits.

search_providers

Search and filter providers to discover environment details and documentation.

get_provider

Fetch detailed information about a specific provider, including environment variables and API documentation.