Skip to main content
POST
/
v3
/
video-agents
Create Video Agent Session
curl --request POST \
  --url https://api.heygen.com/v3/video-agents \
  --header 'Content-Type: application/json' \
  --header 'x-api-key: <api-key>' \
  --data '
{
  "prompt": "<string>",
  "mode": "generate",
  "avatar_id": "<string>",
  "voice_id": "<string>",
  "style_id": "<string>",
  "orientation": "landscape",
  "files": [
    {
      "type": "<string>",
      "url": "<string>"
    }
  ],
  "callback_url": "<string>",
  "callback_id": "<string>",
  "incognito_mode": false,
  "auto_proceed": false,
  "skip_agentic_stop": false
}
'
{
  "data": {
    "session_id": "<string>",
    "status": "generating",
    "created_at": 123,
    "video_id": "v_abc123def456"
  }
}

Authorizations

​
x-api-key
string
header
required

HeyGen API key. Obtain from your HeyGen dashboard.

Body

application/json

Request body for creating a video from a prompt using Video Agent v3.

All configuration is flat (no nested config object). Files use the type-discriminated AssetInput union for flexible asset inputs.

Supports two modes:

  • generate (default): one-shot — auto-proceeds through storyboard, produces one video.
  • chat: multi-turn — may pause for user input on real decisions (e.g. pick a voice), auto-proceeds on confirmations. Allows revisions and follow-up videos.
​
prompt
string
required

The message/prompt for video generation (1-10000 characters)

Required string length: 1 - 10000
​
mode
enum<string>
default:generate

Session mode. 'generate' produces one video (fire-and-forget). 'chat' enables multi-turn interaction — the agent may pause for decisions and allows revisions.

Available options:
generate,
chat
​
avatar_id
string | null

Specific avatar ID to use

​
voice_id
string | null

Specific voice ID to use for narration

​
style_id
string | null

Style ID from GET /v3/video-agents/styles. Applies a curated visual template to the generated video.

​
orientation
enum<string> | null

Video orientation. If not provided, auto-detected from content.

Available options:
landscape,
portrait
​
files
(AssetUrl · object | AssetId · object | AssetBase64 · object)[] | null

Optional file attachments (max 20 files)

Maximum array length: 20

Asset input via publicly accessible HTTPS URL.

​
callback_url
string | null

Webhook URL for completion/failure notifications

​
callback_id
string | null

Optional callback ID included in webhook payload

​
incognito_mode
boolean
default:false

When enabled, disables memory injection and extraction for this session

​
auto_proceed
boolean
default:false

If true, skip interactive review and go straight to video generation (no storyboard approval step)

​
skip_agentic_stop
boolean
default:false

Set true to disable agentic stop (not recommended)

Response

Successful response

​
data
CreateVideoAgentResponse · object

Response from creating a video agent session.

Designed as a session resource for future extensibility — GET /v3/video-agents/{session_id} can return the same shape enriched with more fields (messages, video URL, etc.).