Skip to main content
Not yet. The official model is polling GET /api/v1/jobs/{id}. If you need notifications, run a worker whose job is to poll and emit events into your own system. A webhook format can be added later without breaking existing clients.
The Vidgen API does not send CORS headers, so browsers block reading the response from another domain. Put the API calls on your backend. This is deliberate: an API key must not sit in code that reaches end users.
No limit is published. What matters: usage limits attach to the account, not to the key. Three keys do not mean three concurrency allowances.
Fetch GET /api/v1/models, then mirror the server’s price lookup order: quality:resolution → resolution:duration → duration → resolution → creditsDefault. Prices can change on the admin side, so do not copy the numbers into your code.
Possibly. Links shaped like /api/media-proxy?token=... are valid for 24 hours. Download the file as soon as the status is succeeded and store it in your own storage.
Almost always the cause is retrying without an idempotencyKey. Read Idempotency: the key must be at least 8 characters long and reused on every attempt.
No. Credits are refunded automatically when a job fails, when the cancellation fails to save, or when the queue is full. On a failed job, creditsCharged ends up at 0. What you do pay for is output that was actually produced, even if it does not match your artistic expectations.
Vidgen filters sensitive words, public figure names, and indications of abuse before the request is forwarded. The prompt field in the response is the already-filtered content. If what remains damages your creative intent, rewrite it in neutral language.
Some models are sold without filtering and are recognizable by their name. Using the API does not change the rules: the key holder is responsible for the material submitted, and Vidgen’s terms of service apply on every channel. Read the terms in the dashboard before building a product on top of those models.
Music models do appear in the catalog, but that path is not open to API keys yet. Calling it today ends in failed with credits refunded. Audio generation is still dashboard-only.
Usually 1-5 minutes for short clips, but it depends heavily on the queue. What is guaranteed is not the time but the upper bound: 2 hours, after which the job is marked failed and the credits come back.
The limit of 2 concurrent processes only hangs on Unlimited quota usage. If you pay per credit, you are not subject to it — the consequence is a longer queue. For special needs (a studio with many artists), contact support through the dashboard.
Not yet. What exists are the minimal clients in the Node.js and Python guides, plus the OpenAPI spec so you can generate your own client. Everything uses plain HTTP, so your integration does not depend on third-party packages.
All five endpoints in the API Reference tab are stable at v1. Breaking changes will be announced as a new version, and v1 keeps being served.
Through support in the Vidgen dashboard. Include: job.id, the endpoint, the HTTP status, the error exactly as you received it, the origin you used, and the time of the event. Without internal system names — they never appear in a response.