Generate Videos with PixVerse

What you can do

Use WorldRouter’s native PixVerse-compatible routes to create videos with the official PixVerse pixverse-c1 and pixverse-v6 models.

Before you start

PixVerse generation is asynchronous. Create a video task first, then poll the result route until the generated video URL is ready.

Task routes

• POST /openapi/v2/video/text/generate
• POST /openapi/v2/image/upload
• POST /openapi/v2/media/upload
• POST /openapi/v2/video/img/generate
• POST /openapi/v2/video/transition/generate
• POST /openapi/v2/video/fusion/generate
• POST /openapi/v2/video/extend/generate
• GET /openapi/v2/video/result/{video_id}

Quick start flow

1. 1Create a PixVerse task with POST /openapi/v2/video/text/generate.
2. 2Save the returned Resp.video_id.
3. 3Check the result with GET /openapi/v2/video/result/{video_id} until Resp.status is 1.
4. 4Read Resp.url from the result response and download the generated video from that URL.

Supported request models

These are the request model names you should send in the model field:

Text-to-video

Start here if you want the simplest PixVerse workflow.

This example uses a conservative smoke-test payload that is accepted by the PixVerse upstream API.

curl -X POST "https://inference-api-pre-d80ca3.worldrouter.ai/openapi/v2/video/text/generate" \
-H "Authorization: Bearer your_api_key" \
-H "Content-Type: application/json" \
-H "Ai-trace-id: pixverse-c1-smoke-001" \
-d '{
  "model": "pixverse-c1",
  "prompt": "A small blue cube rotating on a white desk, clean studio lighting",
  "duration": 1,
  "quality": "360p",
  "aspect_ratio": "16:9",
  "generate_audio_switch": false
}'

You can start with this example as-is, then adjust the parts that matter most for your use case:

• model: choose pixverse-c1 or pixverse-v6.
• prompt: describe the subject, style, setting, and motion you want.
• duration: choose the generated video duration supported by PixVerse.
• quality: choose the output quality supported by PixVerse, such as 360p or 720p.
• aspect_ratio: include an explicit aspect ratio such as 16:9.
• generate_audio_switch: set whether PixVerse should generate audio when the selected model and account support it.

Typical create response:

{
"ErrCode": 0,
"ErrMsg": "success",
"Resp": {
  "video_id": 407432230962212,
  "credits": 3,
  "credits_unit": "worldrouter"
}
}

This response confirms that the task was created successfully. Your video is not ready yet. WorldRouter returns Resp.credits in WorldRouter Credits.

Save the returned Resp.video_id, then continue to Check video result.

Check video result

Use the video_id returned from the create response to check whether the video is still running, has succeeded, or has failed.

curl "https://inference-api-pre-d80ca3.worldrouter.ai/openapi/v2/video/result/407432230962212" \
-H "Authorization: Bearer your_api_key" \
-H "Ai-trace-id: pixverse-c1-result-407432230962212"

Typical running response:

{
"ErrCode": 0,
"ErrMsg": "Success",
"Resp": {
  "id": 407432230962212,
  "status": 5
}
}

When Resp.status becomes 1, use Resp.url to download the final video:

{
"ErrCode": 0,
"ErrMsg": "Success",
"Resp": {
  "id": 407432230962212,
  "status": 1,
  "url": "https://media.pixverse.ai/pixverse/mp4/example.mp4",
  "outputWidth": 640,
  "outputHeight": 360,
  "has_audio": false,
  "credits": 3,
  "credits_unit": "worldrouter"
}
}

PixVerse status values are provider-native. The most common values are:

Download the video

After the result response includes Resp.url, download the video from that URL:

curl -L "https://media.pixverse.ai/pixverse/mp4/example.mp4" \
-o pixverse-output.mp4

You can also extract and download the URL in one shell flow:

VIDEO_URL="$(curl -s "https://inference-api-pre-d80ca3.worldrouter.ai/openapi/v2/video/result/407432230962212" \
-H "Authorization: Bearer your_api_key" | jq -r '.Resp.url')"

curl -L "$VIDEO_URL" -o pixverse-output.mp4

Image and transition routes

PixVerse also supports image-guided and multi-input video routes. They use the same authentication, ownership, polling, and result flow:

Send the PixVerse-compatible JSON body for the route you are using, then poll GET /openapi/v2/video/result/{video_id} the same way as text-to-video.

Upload an image

Use this request before image-to-video, transition, or fusion workflows. The returned Resp.img_id is the value to send in later video requests. WorldRouter accepts multipart image upload bodies up to 20 MiB by default. Larger requests return 413 before they are forwarded to PixVerse.

curl -X POST "https://inference-api-pre-d80ca3.worldrouter.ai/openapi/v2/image/upload" \
-H "Authorization: Bearer your_api_key" \
-H "Ai-trace-id: pixverse-image-upload-001" \
-F "image=@/absolute/path/to/reference.png"

Typical upload response:

{
"ErrCode": 0,
"ErrMsg": "success",
"Resp": {
  "img_id": 123456789,
  "img_url": "https://media.pixverse.ai/pixverse/image/example.png"
}
}

Upload video or audio media

Use this request before extending an uploaded external video. The returned Resp.media_id is the value to send as video_media_id in an extend request. WorldRouter accepts multipart media upload bodies up to 200 MiB by default. Larger requests return 413 before they are forwarded to PixVerse.

curl -X POST "https://inference-api-pre-d80ca3.worldrouter.ai/openapi/v2/media/upload" \
-H "Authorization: Bearer your_api_key" \
-H "Ai-trace-id: pixverse-media-upload-001" \
-F "file=@/absolute/path/to/input.mp4"

Typical media upload response:

{
"ErrCode": 0,
"ErrMsg": "success",
"Resp": {
  "media_id": 987654321,
  "media_type": "video",
  "url": "https://media.pixverse.ai/pixverse/mp4/example.mp4"
}
}

Image-to-video request

Use img_id from a PixVerse image upload. Do not send a raw image URL in this field.

{
"model": "pixverse-v6",
"img_id": 123456789,
"prompt": "The product photo slowly rotates on a glossy tabletop",
"duration": 5,
"quality": "540p",
"aspect_ratio": "16:9",
"motion_mode": "normal",
"generate_audio_switch": false
}

Transition request

Upload both images first, then use the returned image IDs for the first and last frames.

{
"model": "pixverse-v6",
"first_frame_img": 123456789,
"last_frame_img": 987654321,
"prompt": "A smooth cinematic transition from morning to evening",
"duration": 5,
"quality": "540p",
"aspect_ratio": "16:9",
"motion_mode": "normal",
"generate_audio_switch": false
}

Fusion request

Use image_references to bind uploaded images to names you mention in the prompt.

{
"model": "pixverse-v6",
"image_references": [
  {
    "type": "subject",
    "img_id": 123456789,
    "ref_name": "product"
  },
  {
    "type": "background",
    "img_id": 987654321,
    "ref_name": "studio"
  }
],
"prompt": "Place @product on @studio with a slow dolly-in camera move",
"duration": 5,
"quality": "540p",
"aspect_ratio": "16:9",
"generate_audio_switch": false
}

Extend request

Use source_video_id from an earlier PixVerse generation response, or upload an external video first and use the returned media_id as video_media_id.

{
"model": "pixverse-v6",
"source_video_id": 407432230962212,
"prompt": "Continue the same camera motion for a few more seconds",
"duration": 5,
"quality": "540p",
"generate_audio_switch": false
}

{
"model": "pixverse-v6",
"video_media_id": 987654321,
"prompt": "Continue the same camera motion for a few more seconds",
"duration": 5,
"quality": "540p",
"generate_audio_switch": false
}

Notes and limits

• PixVerse requires a team-scoped key with enough available credits. Low balance returns 402 with a PixVerse balance error.
• Pending PixVerse jobs per user are capped. Too many running tasks return 429.
• Result reads are owner-scoped. You can only poll videos created by the same authorized owner context.
• WorldRouter records successful terminal PixVerse results for usage billing when Resp.status is 1.
• Keep aspect_ratio explicit in create requests. Missing or unsupported PixVerse fields may return 400013 Invalid field type or value from the upstream API.

Troubleshooting

See also

• Quickstart: API key and base URL setup.
• Models: current model catalog and pricing.
• Seedance video guide: async video generation with Seedance.
• API reference: chat-completions docs for text models.