Runway MCP Server
November 21, 2026
Table of contents
- useapi.net vs Runway’s own MCP servers
- Endpoint and authorization
- Setup
- Use it in the Claude app or ChatGPT
- Tools
- Long-running work
- Example prompts
- Limits and notes
The useapi.net Runway MCP server lets an AI assistant such as the Claude app, ChatGPT, Claude Code, Cursor or VS Code Copilot make videos and images with Runway on your own Runway accounts. It is a hosted Model Context Protocol server with 25 tools: generate video with any of the 25 video models Runway hosts, among them Gen-4.5, Seedance 2.5, Kling O3, Veo 3.1, Wan 3.0 and MiniMax H3, edit a video with Aleph, animate a character with Act-Two, swap the voice of a video or lip sync a face, generate images with 13 models or Frames, upscale any video to 4K with Topaz, turn an image into a prompt for free, and manage your uploads and task history. Each tool calls the Runway API, so the same rules and limits apply. Nothing needs to be installed or hosted.
Is there a Runway MCP server? Yes. useapi.net, a third-party service not affiliated with Runway, hosts one at https://api.useapi.net/v1/runwayml/mcp with 25 tools that spread tasks across several of your Runway accounts, with Explore Mode on every generation tool. The same token runs your REST scripts and webhooks on your Runway subscription credits, which Runway’s developer API can’t spend, so that is only available through useapi.net. Runway’s own hosted MCP signs in to one account (comparison).
Prefer REST? Code: github.com/useapi/runway-api
See all useapi.net MCP servers on Model Matrix & MCP.
useapi.net vs Runway’s own MCP servers
Runway runs its own hosted Runway MCP for ChatGPT, Claude, Cursor and other apps: you sign in with your Runway app account and it spends that account’s credits. Its tool list is not published, and tools/list asks for a sign-in. The Runway Dev MCP is a separate server for developer accounts, for tasks, model routers and docs, and Runway’s open-source runway-api-mcp-server was archived on September 30, 2026. What the useapi.net server adds (as of November 21, 2026):
| useapi.net Runway MCP | Runway MCP (mcp.runwayml.com) | |
|---|---|---|
Several Runway accounts in one connection, each task on an account with a free slot (maxJobs) | Yes | One Runway account, the one you sign in with |
The same token runs your own scripts on the same subscription credits, with webhooks (replyUrl) | Yes, through the REST API | No, scripts need Runway’s developer API, with separate credits at $0.01 each |
| Every tool, model and parameter documented before you connect | Yes, Tools below | Tool list not published |
| Video models documented | 25, among them Wan 3.0, MiniMax H3, HappyHorse 1.0, Grok Imagine 1.5, FLUX 3 and Kling O3 | runway.com/mcp names Gen-4.5, Seedance 2.5, Kling 3.0, Veo 3.1 and Gen-4 Turbo “and more” |
Explore Mode (exploreMode) on every generation tool | Yes | Not mentioned on runway.com/mcp |
| Tools for Act-Two, voice swap, lip sync, Aleph edits and Topaz 4K upscales | Yes | Not mentioned on runway.com/mcp |
| Upload, list and delete assets, list and delete tasks | Yes | Not mentioned on runway.com/mcp |
useapi.net is a third-party service, not affiliated with Runway.
Endpoint and authorization
https://api.useapi.net/v1/runwayml/mcp
- Transport: Streamable HTTP.
- Authorization: your useapi.net API token in the
Authorization: Bearer …header, see Setup useapi.net. The Claude app and ChatGPT sign in with OAuth instead, see Use it in the Claude app or ChatGPT. - You need at least one Runway account added with Setup Runway. Adding and removing accounts is not available over MCP: use POST accounts or the setup page.
Setup
Replace YOUR_API_TOKEN with your useapi.net API token.
Claude Code
claude mcp add --transport http useapi-runwayml https://api.useapi.net/v1/runwayml/mcp --header "Authorization: Bearer YOUR_API_TOKEN"
Cursor
In ~/.cursor/mcp.json:
{
"mcpServers": {
"useapi-runwayml": {
"url": "https://api.useapi.net/v1/runwayml/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_TOKEN"
}
}
}
}
VS Code
In .vscode/mcp.json:
{
"servers": {
"useapi-runwayml": {
"type": "http",
"url": "https://api.useapi.net/v1/runwayml/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_TOKEN"
}
}
}
}
Windsurf and other clients
Any client that supports remote MCP servers over HTTP with custom headers works. The usual JSON shape:
{
"mcpServers": {
"useapi-runwayml": {
"serverUrl": "https://api.useapi.net/v1/runwayml/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_TOKEN"
}
}
}
}
Some clients name the address field url instead of serverUrl.
Test it with curl
curl -s https://api.useapi.net/v1/runwayml/mcp \
-H "Authorization: Bearer YOUR_API_TOKEN" \
-H "Content-Type: application/json" \
-H "Accept: application/json, text/event-stream" \
-d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-06-18","capabilities":{},"clientInfo":{"name":"curl","version":"1"}}}'
curl -s https://api.useapi.net/v1/runwayml/mcp \
-H "Authorization: Bearer YOUR_API_TOKEN" \
-H "Content-Type: application/json" \
-H "Accept: application/json, text/event-stream" \
-d '{"jsonrpc":"2.0","id":2,"method":"tools/list"}'
The first call answers with the server’s name useapi-runwayml and its instructions for the assistant, the second with the 25 tools.
Use it in the Claude app or ChatGPT
The Claude app (claude.ai and Claude Desktop) and ChatGPT can add this server as a custom connector. They sign in with OAuth: you give the app the URL, it opens a useapi.net page, and you paste your useapi.net API token there once. Both work on their Free plans. The API-token setups above for Claude Code, Cursor, VS Code and other clients keep working as before.
https://api.useapi.net/v1/runwayml/mcp
Claude app
- Open Settings → Connectors and click Add custom connector.
- Enter a name, for example
useapi Runway, and the URL above, then click Add. - Click Connect. Claude opens the useapi.net sign-in page.
- Paste your useapi.net API token and click Allow.
Claude asks before each tool call unless you choose Always allow. The Free plan allows one custom connector.
ChatGPT
- Open Plugins, click Add and choose Add custom MCP server.
- Enter a name and the URL above, and set Authentication to OAuth.
- Tick ChatGPT’s “I understand and want to continue” box, then click Create as a plugin.
- Click Connect. ChatGPT opens the useapi.net sign-in page.
- Paste your useapi.net API token and click Allow.
ChatGPT asks you to confirm each tool call that creates or changes something. The Free plan works without developer mode (as of October 25, 2026).
How the sign-in works
- Your API token is in your welcome email, see Setup useapi.net. The useapi.net page links there too.
- Each connection is for one MCP server. Connect each server you want to use on its own.
- The app gets an access token that lasts 1 hour and renews it on its own. A connection you don’t use for 90 days lapses, and you connect again.
- Resetting your API token disconnects every app connected with it.
- Results come back as public links, so they open right in the app.
Tools
Every generation is a Runway task with a taskId, read with runway_get_task. Reference images, videos and audio are assets: upload your own with runway_upload_asset, or reuse the assetId of an earlier upload or of a finished task. A task that uses assets runs on the Runway account that holds them. Without assets, runway_create_video, runway_create_image and the other tools that take email run on one of your accounts with a free slot, or on the account you name with email. Every generation tool also takes exploreMode, wait, and a webhook with replyUrl and replyRef.
Videos
| Tool | What it does | Main parameters |
|---|---|---|
runway_create_video | Generates a video with any of the 25 video models on Runway: Seedance 2.5, 2.0, 2.0 Fast and 2.0 Mini, Wan 3.0, 3.0 Prime, 2.6 Flash and 2.2 Animate, MiniMax H3 and H3 Max, FLUX 3, Grok Imagine 1.5, HappyHorse 1.0, Kling O3, 3.0, 3.0 Motion Control and 2.6, Veo 3.1, Gen-4.5, Gen-4 and Gen-4 Turbo. From a prompt, start and end frames, reference images, videos and audio, multi-shot prompts on Kling, and drafts on Seedance 2.5. | model, text_prompt, duration, aspect_ratio, resolution, audio, video_mode, draft, startFrameAssetId, endFrameAssetId, imageAssetId1 to imageAssetId30, videoAssetId to videoAssetId10, audioAssetId1 to audioAssetId10, multishot_prompt_1 to _6, email, maxJobs |
runway_create_gen4_5 | Generates a Gen-4.5 video through its own route: text-to-video in 16:9, or image-to-video from a first frame, 5, 8 or 10 seconds. runway_create_video runs Gen-4.5 at any length from 2 to 10 seconds with an audio switch. | text_prompt, firstImage_assetId, aspect_ratio, seconds, email, maxJobs |
runway_create_gen4 | Generates a Gen-4 video from a first-frame image and an optional prompt, 5 or 10 seconds. | firstImage_assetId, text_prompt, aspect_ratio, seconds |
runway_create_gen4_turbo | Generates a Gen-4 Turbo video from a first-frame image and an optional prompt, 5 or 10 seconds. | firstImage_assetId, text_prompt, aspect_ratio, seconds |
runway_edit_video_aleph | Changes an existing video with Gen-4 Aleph from a text prompt, optionally guided by an image. | video_assetId, text_prompt, image_assetId |
runway_act_two | Animates a character image or video like the person in a driving video of 3 to 30 seconds, with body control and expression intensity. | driving_assetId, character_assetId, aspect_ratio, body_control, expression_intensity |
runway_act_two_voice | Replaces the voice in a video of 3 to 30 seconds with one of Runway’s voices, keeping the picture. | video_assetId, voiceId |
runway_lipsync | Makes the face in an image or video speak an audio clip, or your text read by one of Runway’s voices. | image_assetId or video_assetId, audio_assetId or voiceId with voice_text, model_id |
runway_list_lipsync_voices | Lists the voices for lip sync and voice swap, each with its voiceId, name, labels and a sample MP3. | email |
The parameters match POST videos/create, POST gen4_5/create, POST gen4/create, POST gen4turbo/create, POST gen4/video, POST gen4/act-two, POST gen4/act-two-voice, POST lipsync/create and GET lipsync/voices. Durations, resolutions and references of each model are on POST videos/create.
Images
| Tool | What it does | Main parameters |
|---|---|---|
runway_create_image | Generates 1 or 4 images with any of the 13 image models on Runway: FLUX 3, FLUX.2 Max and Klein, Nano Banana, 2, 2 Lite and Pro, GPT Image 2, 1.5 and 1 Mini, Seedream 5, Gen-4 and Gen-4 Turbo, from a prompt and up to 14 reference images. The first image comes back as an image the assistant can see. | model, text_prompt, aspect_ratio, resolution, num_images, imageAssetId1 to imageAssetId14, quality, email |
runway_create_frames | Generates 1 or 4 images at 1080p with Runway Frames, with up to 3 reference images cited as @IMG_1 to @IMG_3, a style preset and an aesthetic range. | text_prompt, imageAssetId1 to imageAssetId3, style, aspect_ratio, diversity, num_images |
runway_describe_image | Turns an image into a detailed description to edit and use as a prompt. It answers at once, with no task, and is free and unlimited on every Runway account, the free plan too. | image_assetId |
The parameters match POST images/create, POST frames/create and GET frames/describe.
4K upscale
| Tool | What it does | Main parameters |
|---|---|---|
runway_upscale_video | Upscales any video of up to 40 seconds to 4K with Topaz, keeping its aspect ratio: a video from any Runway task or your own upload. | videoAssetId |
runway_upscale_gen4 | Upscales a Gen-4, Gen-4 Turbo, Gen-3 Alpha or Gen-3 Alpha Turbo video to 4K with Runway’s own upscaler. | assetId |
The parameters match POST videos/upscale and POST gen4/upscale.
Assets
| Tool | What it does | Main parameters |
|---|---|---|
runway_upload_asset | Uploads your own image, video or audio file, up to 50 MB, from a public https URL or sent as contentBase64, to a Runway account’s assets. Returns its assetId, URL, name, type and size. | name, url or contentBase64, mimeType, email |
runway_list_assets | Lists one account’s assets, uploads and generated files alike, those from the Runway web app too, with each one’s URL, previews, size, media details and the taskId that made it. | email, mediaType, offset, limit |
runway_get_asset | Reads one asset by its assetId. | assetId |
runway_delete_asset | Deletes one asset from its Runway account. It cannot be undone. | assetId |
The parameters match POST assets, GET assets, GET assets/assetId and DEL assets/assetId. Files larger than 50 MB go to POST assets directly.
Tasks
| Tool | What it does | Main parameters |
|---|---|---|
runway_get_task | Reads one task, and with wait: true waits for it to finish. A finished task comes with its files (url and assetId), a failed one with its error. | taskId, wait |
runway_list_tasks | Lists one account’s task history, tasks started in the Runway web app included, with each one’s status, progress and finished files. statuses filters it, for example to the tasks still running. | email, statuses, offset, limit |
runway_delete_task | Deletes one task from its account’s task history. It cannot be undone. | taskId |
runway_list_running_tasks | Lists the tasks the API is tracking now, started through it and not yet finished. | |
runway_cancel_task | Stops tracking a running task and frees its slot on the account. It does not cancel the task at Runway: the generation carries on and uses the account’s credits. It cannot be undone. | taskId |
The parameters match GET tasks/taskId, GET tasks, DEL tasks/taskId, GET scheduler and DEL scheduler/taskId.
Accounts and credits
| Tool | What it does | Main parameters |
|---|---|---|
runway_list_accounts | Lists your Runway accounts as stored: email, maxJobs, the paid workspace an account bills against, session times and any stored account error, without calling Runway. | email |
runway_get_features | Reads an account live from Runway: its plan credits and how many are used, its storage allowance and the workspaces it belongs to with their plans. | email |
The parameters match GET accounts, GET accounts/email and GET features.
The full input schema of every tool comes from tools/list.
Long-running work
MCP clients usually give up on a tool call after about a minute, so a tool waits up to about 50 seconds:
- A quick result can come back from the create tool itself.
- Otherwise the tool returns the
taskId, andrunway_get_taskwithwait: truepicks it up. Ask the assistant to keep checking until the task isSUCCEEDEDorFAILED. wait: falseon a create tool returns thetaskIdat once.
Clients that ask for progress notifications get one about every 10 seconds while a tool waits.
Videos and images come back as Runway links, each with an assetId that works as a reference in the next task on the same account.
Example prompts
- “Make a 10-second Gen-4.5 video of a paper boat drifting down a rainy street at night.”
- “Upload this photo and make a Kling O3 video where it slowly comes to life, in 16:9.”
- “Generate 4 Nano Banana Pro images of a cozy mountain cabin in winter.”
- “Restyle this clip with Aleph as a hand-drawn watercolor.”
- “Animate my character image with Act-Two, using this recording of me as the performance.”
- “Make the face in this portrait say ‘Welcome to the show’ in one of Runway’s voices.”
- “Upscale the last video to 4K.”
- “How many credits are left on my Runway account?”
Limits and notes
- The tools use your Runway accounts and the same rules as the REST API, such as
maxJobs(1 to 10 tasks at once per account). Generations spend the account’s Runway credits, see Pricing, andrunway_get_featuresshows what is left. - Accounts on Runway’s Unlimited plan can set
exploreModeto make videos, images and 4K upscales without spending credits, at lower priority. It works on every model except Veo 3.1, Wan 3.0 and 3.0 Prime, MiniMax H3 and H3 Max, and FLUX 3 video and images, and Seedance runs at 480p or 720p in it. - A 429 means your accounts are busy or Runway’s own queue is full: check
runway_list_running_tasksand retry later. A 596 means the account needs attention: fix it on the setup page. - A client may send up to 20 JSON-RPC requests in one batch.
- If the client sends an
MCP-Protocol-Versionheader, it must be one of the supported versions:2025-11-25,2025-06-18,2025-03-26or2024-11-05.