Runway MCP server for your AI

Make a video, image or audio and get the result: the normal way to generate.

See which models Runway has for each job, and exactly which fields one takes, because ratio, duration and other fields change from model to model.

Start making a video from an image and a prompt.

Start making images from a prompt, optionally guided by reference images.

Start making a video from a text prompt.

Create an Avatar from a reference image, a personality and a voice.

Save a new Model Router.

Get one Avatar: status, personality, voice and attached documents.

List past live conversations, newest first.

List this project's Avatars (reusable characters for live conversation and avatar videos), with status, voice and id.

List saved Model Routers, each a saved rule for choosing a model by cost, speed or quality.

See which model a saved Model Router would pick for a request, and the estimated credits, without generating or spending anything.

Start a video of an Avatar speaking text or audio, with no live conversation.

Start a video through a saved router, which picks the model.

Start making a product ad video from product images and an idea.

Create a custom voice, from a text description or by cloning an audio sample of 10 seconds to 5 minutes.

Get one published workflow version, with inputs: the node ids and keys you can set when running it.

Check a workflow run once: status, output (node id to links) and node_errors.

List this project's custom voices (designed or cloned), with status and preview link.

List workflows published to this project, each with its versions newest first.

Run a published workflow and get its results.

Wait for a workflow run that is still going, and get its output when it ends.

Add a knowledge document for Avatars to answer from: a name and up to 200,000 characters of text.

Get credits spent per day and per model for this project.

Get the credit balance, usage tier and per-model limits of the project this key belongs to, and today's use per model.

List knowledge documents with name, type and which Avatars use them.

Stop a task that is still going, or delete a finished one along with its output.

Open a live conversation with an Avatar.

Check a live conversation: status, when it expires, and failure.

Check one task once: status (PENDING, THROTTLED, RUNNING, SUCCEEDED, FAILED, CANCELLED), progress, output_urls, credits and failure.

Wait for a Runway task that is still going, and get its output when it ends.

Put a file on Runway so a generation can use it, when it has no public link or is too big to send inline.

Start applying a filmed performance from a reference video to a character in an image or video.

Start upscaling an image to a higher resolution.

Start making a sound effect from a description.

Start re-speaking an audio or video clip in another voice.

Start turning text into spoken audio.

Start making audio that fits a video.

Start converting a standard video to HDR or a mastering format.

Start restyling or editing a video with a prompt.

Start upscaling a video, or changing its frame rate, depending on the model.

Start dubbing a clip into another language.

Start separating speech from background sound in a clip.

Delete an Avatar.

Delete a conversation and its transcript and recording.

Delete a Model Router.

Get one conversation with its transcript and recording link.

Get conversation totals for a date range: duration, session counts, average length and a day-by-day breakdown in UTC.

Get one Model Router and its settings.

List a router's recorded decisions: which model it picked and why.

Start audio through a saved router, which picks the model.

Start an image through a saved router, which picks the model.

Change an Avatar's name, image, personality, voice or documents.

Change a Model Router's name, description or settings.

Start localizing an ad image into another language.

Start making a marketing stock image from a prompt.

Start making a video of several shots, from one prompt (mode auto) or your own shot list (mode custom).

Start making product campaign images from a product image and a prompt.

Start swapping a product in a reference video for another product.

Start making a creator-style product video from a character image and a product image.

Delete a custom voice.

Get one custom voice: status, preview link and failure reason.

Hear a sample of a voice described in words before creating it.

Start a published workflow.

Rename a custom voice or change its description.

Delete a knowledge document.

Get one knowledge document, with its text.

Get one audit log entry from a linked Runway web app workspace.

List audit log entries from linked Runway web app workspaces, newest first.

List per-generation credit use from Runway web app workspaces linked to this project, newest first.

Change a knowledge document's name or text.

End a live conversation that is still active.

Low level first step of an upload: answers uploadUrl, fields and runwayUri, and you still have to send the file there yourself.