VidLabDocs
Copy for agent
llms.txtEvery page and model as links an agent can follow.Agent setup guideThe skill, your key and what to ask for.
StatusSupportGet API key

Vidra 3.0

VidLabvidra-3.0

Our own video model: a prompt with up to 30 reference images, 10 reference videos and 10 reference sounds, or a first and last frame · 480p to 4K · 4 to 30 s · native audio

Video generation
POSThttps://vidlab.ai/v1/generations

Send "model": "vidra-3.0" and the fields below. The call answers 202 at once with the generation's id; poll it for the file. Billed per second of output, only when it succeeds.

Request
curl -X POST https://vidlab.ai/v1/generations \
  -H "Authorization: Bearer $VIDLAB_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: $(uuidgen)" \
  -d '{
  "model": "vidra-3.0",
  "input": {
    "prompt": "A slow push-in on the product in @Image 1",
    "resolution": "480p",
    "aspect_ratio": "adaptive",
    "duration": 5,
    "generate_audio": true,
    "images": [
      "https://example.com/reference-1.jpg"
    ],
    "image_mode": "reference"
  }
}'
Response
{
  "data": {
    "id": "gen_4Fq9xT2mLr7Kd1sV8bN3pZ0c",
    "object": "generation",
    "status": "queued",
    "model": "vidra-3.0",
    "variant": "480p",
    "units": 5,
    "price_usd": 0.5,
    "outputs": [],
    "created_at": "2026-11-01T18:04:11.000Z",
    "status_url": "/v1/generations/gen_4Fq9xT2mLr7Kd1sV8bN3pZ0c",
    "result_url": "/v1/jobs/gen_4Fq9xT2mLr7Kd1sV8bN3pZ0c/result"
  }
}

Headers

HeaderTypeRequiredValue
AuthorizationstringrequiredBearer vidlab_sk_..., your API key from the dashboard.
Content-Typestringrequiredapplication/json
Idempotency-Keystringoptional8 to 100 letters, digits or _ : . - (a UUID is fine). A retry with the same key never starts or charges twice.

No key yet? Make one under API keys. Keep it on your server, never in a browser or an app.

Body

FieldTypeRequiredDefaultAllowed
modelstringrequiredNone
Must be vidra-3.0
inputobjectrequiredNone
The fields below.
max_cost_usdnumberoptionalNone
Dollars
The most you will pay. Over it, the call answers 422 and nothing is held. Submit a generation
max_wait_snumberoptionalNone
Seconds
The longest you will wait.

Input

The fields of input for video generation. A field the model does not take is refused, never ignored. Files are https links or base64 data URLs; upload a file from your machine first.

FieldTypeRequiredDefaultAllowed
promptstringrequiredNone
Up to 20,000 characters
What to make.
resolutionstringoptional480p
One of: 480p, 720p, 1080p, 2k, 4k
Output resolution.
aspect_ratiostringoptionalFollows your input
One of: 16:9, 9:16, 1:1, 4:3, 3:4, 21:9, adaptive
Shape of the output.
durationnumberoptional4
4 to 30 seconds
Seconds of output.
generate_audiobooleanoptionaltrue
One of: true, false
Ask for sound where the model makes it.
imagesarray of filesoptionalNone
Up to 30 files: JPEG, PNG or WEBP, up to 20 MB
Reference images. A first frame or main reference goes first.
image_modestringoptionalreference
One of: first_frame, reference
The page sends reference: every image is a reference. first_frame opens the video on images[0] (no videos or audio then).
videosarray of filesoptionalNone
Up to 10 files: MP4 or MOV, up to 100 MB, 2 to 15 s each, 30 s together (0.4 MP to 8.3 MP)
More reference clips, after video. Their seconds are billed too. Billing
audiosarray of filesoptionalNone
Up to 10 files: MP3 or WAV, up to 15 MB, 2 to 15 s each, 30 s together
Reference sounds: music, a voice or a beat to follow.
seedintegeroptionalNone
A whole number
Repeatable results where the model supports it.
extraobjectoptionalNone
Up to 8 KB as JSON
To price the call on the reference clips' real length, send video_seconds: all of them together, in whole seconds (the page does this for you). Billing

Pricing

Billed per second of output. A failed generation is free. The example costs $0.50 (480p, 5 seconds). Check any request for free with POST /v1/generations/quote.

OptionPriceBilled
480p 480p$0.10per second of output
720p 720p$0.23per second of output
1080p 1080p$0.57per second of output
2k 2k$1.40per second of output
4k 4k$1.54per second of output
480p+video_input 480p, with a reference video$0.07per second of output
720p+video_input 720p, with a reference video$0.14per second of output
1080p+video_input 1080p, with a reference video$0.34per second of output
2k+video_input 2k, with a reference video$0.82per second of output
4k+video_input 4k, with a reference video$0.91per second of output

Response

202 with the generation (status: "queued"), its price, and where to poll it. The fields are on Generation object.

Get the result

GEThttps://vidlab.ai/v1/generations/{id}

Poll every few seconds until status is succeeded or failed. outputs[].url is the file, on our storage; download it with your key from GET /v1/generations/{id}/content. See Get the result.

curl https://vidlab.ai/v1/generations/gen_4Fq9xT2mLr7Kd1sV8bN3pZ0c?wait=60 \
  -H "Authorization: Bearer $VIDLAB_API_KEY"

Errors

Every error is { "error": { "message", "type", "code" } } and nothing is held. All codes are on Errors.

StatusCodeWhen
400invalid_inputA field this model does not take, a value that is not allowed, or a required field left out. The message names it.
401unauthorizedThe key is missing, unknown or revoked.
422max_cost_exceededNo route can run the call for max_cost_usd or less. Nothing is held.
429api_wallet_exhaustedThe balance has no room for this price. Add credit.
429rate_limitedThis account started 20 generations in the last 10 seconds. Not queued: wait for Retry-After, then send it again.
503api_wallet_reconciliation_requiredThe balance is held for a billing review. Contact support; do not retry until it is cleared.