MCP server: One MCP server.

ModelRunner

@modelrunnerMCP server · External

Run 100+ AI models — image, video, audio, 3D — through one API with pay-per-use billing.

Freeno charge per call

One address for any MCP client. You sign in at mcp.modelrunner.run once, after adding it.

What it does

  1. Reads
  2. Writes
  3. High impact

At most: moves money, runs code or commands, deletes data, or sends messages for you.

What it costs

Free

Odel charges nothing per call. The publisher may charge for its own service.

What it asks for at sign-in

Access toReadChange
McpNot stated: mcp

From the sign-in at mcp.modelrunner.run, as the endpoint publishes it. You approve it there.

Tools

None listed · checked 10 Oct 2026

Odel could not list its tools without signing in.

Try it in the Inspector
README3 min read

ModelRunner MCP Server

Official Model Context Protocol (MCP) server for ModelRunner — run 100+ AI models across image, video, audio, and 3D generation from Claude, Cursor, and any MCP client.

  • Endpoint: https://mcp.modelrunner.run/mcp (remote server, Streamable HTTP)
  • Auth: OAuth 2.1 — your browser authorizes once; no API key is pasted into config
  • Registry name: ai.modelrunner/mcp
  • Setup guide & full tool reference: modelrunner.ai/mcp

Unlike single-vendor MCP servers, one connection exposes the entire ModelRunner catalog — models from multiple AI providers behind one endpoint and one pay-per-use billing account. Your assistant can search the catalog, inspect a model's input schema, run inference, and get results back as hosted URLs directly in the conversation.

Maintaining an open-source project? ModelRunner gives active open-source projects free monthly API credits — apply at modelrunner.ai/oss-program.

Quick start

Claude Code

bash
claude mcp add --transport http modelrunner https://mcp.modelrunner.run/mcp

Cursor / VS Code (native remote URL)

json
{
  "mcpServers": {
    "modelrunner": { "url": "https://mcp.modelrunner.run/mcp" }
  }
}

Claude Desktop or any stdio-only client (via mcp-remote)

json
{
  "mcpServers": {
    "modelrunner": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://mcp.modelrunner.run/mcp"]
    }
  }
}

On first use your AI assistant opens a browser window to sign in to ModelRunner — that's the whole setup.

Tools

GroupTools
Discover & inspectsearch, list_models, list_wrappers, recommended_models, recommended_wrappers, get_model, get_wrapper, get_model_raw_schema, get_wrapper_raw_schema, get_wrapper_prompt_template, get_wrapper_field_mappings
Run & fetchrun_model, wait_for_request, get_request, list_my_requests
Upload filescreate_upload_url, upload_file
Author wrapperswrapper_authoring_guide, create_wrapper, patch_wrapper, preview_wrapper, list_my_wrappers, delete_wrapper

"Wrappers" are reusable, simplified interfaces you can build over base models — a few friendly inputs, a prompt template, and baked-in settings — and they run through the same run_model-style tools.

Models & capabilities

The catalog covers text-to-image, image-to-image, text-to-video, image-to-video, video-to-video, music generation, speech-to-text, and image-to-3D, including models like Kling, HiDream, Hunyuan Image, Stable Audio, and Rodin. Browse by capability:

Full catalog: modelrunner.ai/models

Pricing

Free to connect. Model runs are pay-per-use from prepaid credits — no subscription. Every model's exact price (per image, per second of video, per megapixel, …) is shown on its model page and on the pricing page.

About this repository

This repository documents the hosted ModelRunner MCP server. The server itself is first-party, run by ModelRunner as a remote MCP endpoint — there is nothing to install or self-host; the configs above are the entire integration.

License

MIT

Similar modules

Browse LLM access & routing