Core Guides
FAQ
What to change when a TokenLab request is rejected
Short answers for the failures that stop a first integration. Each answer names the field that was rejected and the page that documents the fix.
A video request returns 400
POST /v1/videos/generations rejects a duration, seconds, resolution, or operation value that the selected model does not support. The error body lists the values that work, so read it before changing the request:
allowed_durations: accepteddurationandsecondsvalues, in seconds.allowed_resolutions: acceptedresolutionvalues.supported_operations: acceptedoperationnames.recommended_request: one request that uses an allowed value; it is absent when the operation needs an input only you can supply, such as an image forimage-to-video, andrequired_selectorsnames that input instead.
The lists belong to the selected model and to the operation you sent, so values from another model do not apply. See Video generation and Create Video.
A Gemini file upload does not work
Two paths need no upload step: send the bytes inline, as a data:image/png;base64,... data URL in the OpenAI-compatible image_url.url field or as raw Base64 (no data: prefix) in the Gemini native inlineData.data field, or send a public https URL in fileData.fileUri. When you need one file referenced by several requests, upload the bytes yourself to POST /upload/v1beta/files with your TokenLab key and use the returned file_uri. The Google Gen AI SDK's own upload helper is not served by the TokenLab host and returns 404, so retrying it cannot succeed. See Upload Gemini file and Gemini Files and cached content.
401 or 402
| Status | Code | What to do |
|---|---|---|
401 | invalid_api_key | Send the key as Authorization: Bearer sk-.... A revoked or inactive key keeps failing until the Workspace creates a new one. |
402 | insufficient_balance | The balance does not cover the request: add funds, reduce the request, or choose a lower-priced model. |
See Authentication, Billing and pricing, and Handle API errors.
Retry without creating two tasks
Send an Idempotency-Key with POST /v1/videos/generations. After a timeout, send the same request with the same key to get the original task instead of a second one. See Handle API errors.