Skip to main content
Tavus Video Generation is the same async, file-output product described in Video overview: you call the Video API and poll until a downloadable video is ready. It is not the Conversational Video Interface (CVI) - there is no live room, Daily room, or PAL-driven real-time session here.

Prerequisites

Before starting, ensure you have:
  • A replica_id for the video (this field name is unchanged on the Video API; the value is a Face ID from Stock faces or your own trained face). For your own likeness, see Face overview and training docs.
  • Script in text form, or audio as .mp3 or .wav (hosted at a URL you can pass for the audio path).
At a glance
  • Create: POST https://tavusapi.com/v2/videos with replica_id plus either script (text) or audio_url (and optional callback_url).
  • Status: GET https://tavusapi.com/v2/videos/{video_id} - Get Video.
  • While generating: status is generating; when ready, download_url / hosted_url populate.

Generate a video

1

Step 1: Generate Your Video

Use the following request to generate a video:
You can also customize the video background to suit your needs. See the Background Customizations article for more details.
2

Step 2: Check Video Generation Status

You can monitor generation using the Get Video endpoint:
Replace <api_key> with your actual API key (see Step 1). Replace <video_id> with the ID returned when you created the video.
If the video is still being generated, the response will include a status field set to generating.
Once the video is fully generated, the response will return a status field set to ready.
3

Step 3: Accessing Your Video

Once generated, videos can be:
  • Streamed or Downloaded: Generated video is hosted on a shareable URL. If a callback is set, a download link is returned when the video generated.
  • Embedded or Shared: Use the provided links to distribute your videos across social media, internal tools, or customer platforms.