Skip to main content
GET
cURL

Overview

Check the status of an asynchronous video generation job using a single, provider-agnostic endpoint. This endpoint accepts NanoGPT job IDs (vid_...) and legacy provider request IDs. No model parameter is required.

Query parameters

*Send either requestId or runId.

Authentication

Only the account that created a job can read its status or its event stream: send that account’s API key (x-api-key) or session. A job created by another account answers 404, exactly like one that doesn’t exist, so the response never confirms that someone else’s job exists. Three exceptions keep the job ID enough on its own: jobs paid anonymously with x402; the statusUrl (and eventsUrl) returned when some private models are submitted, which carry their own access parameters and work for 24 hours; and older jobs NanoGPT has no job record for, whose status is still read by ID alone, as before.

Output URLs

When a video is served from NanoGPT’s content route (/api/generate-video/content?...), a response to the account that created the job (or for an x402 job, or through a statusUrl) gives its URL expires and sig parameters. Such a URL works without an API key for 24 hours, so you can hand it to a player or a download tool. After that, poll the status endpoint again for a fresh URL, or fetch the content route with the creating account’s API key. Anyone else reading an older job by its ID alone gets these URL fields empty: NanoGPT never hands out an unsigned content URL. NanoGPT keeps a video it stores on its servers for about a day after storing it, so download the ones you want to keep; for most stored videos output.expiresAt gives that date. URLs hosted by the model’s provider are returned as the provider gives them and can expire sooner. In the NanoGPT web app, new generated videos are saved in the browser where they finish when storage space is available. If a video cannot be saved, the app asks you to download it. Clearing browser data removes these saved videos, so download the ones you want to keep for good. API clients must download and store their own copies; this endpoint does not save videos on their devices.

Usage

Status values

  • IN_QUEUE: Request queued
  • IN_PROGRESS: Generation in progress
  • COMPLETED: Video ready
  • FAILED: Generation failed
  • CANCELED: Request canceled

Response examples

In progress

Completed

Failed

Notes:
  • Terminal results are cached in video_jobs for faster subsequent status checks.

Authorizations

x-api-key
string
header
required

Query Parameters

requestId
string
required

The unique request ID returned from any video generation endpoint

Response

Unified video generation status

requestId
string
required

The unique request ID for the video generation

status
enum<string>
required

Current normalized status of the video generation

Available options:
queued,
processing,
completed,
failed,
cancelled,
unknown
videoUrl
string | null
required

URL to the generated video (available when status is 'completed')

error
string | null
required

Error message if the generation failed

createdAt
string | null
required

ISO timestamp when the request was created

completedAt
string | null
required

ISO timestamp when the generation completed

seed
integer

The seed NanoGPT sent to the provider. Returned on completed results from models with returned seeds when the account saves request context (the default); omitted otherwise. Identical output is not guaranteed when the seed is reused.

progress
number | null

Generation progress as a percentage (0-100), if available

estimatedTimeRemaining
number | null

Estimated time remaining in seconds, if available