> ## Documentation Index
> Fetch the complete documentation index at: https://docs.orvado.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# FAQ

> Short answers to common partner questions

> **Do I need to change my code to switch from OpenAI?**
>
> Usually not. Set the base URL to `https://orvado.dev/v1`, use your `td-` key and pick a model ID from [Models and pricing](/models-and-pricing).

> **How do I top up my balance?**
>
> Pay as agreed with your Orvado manager. The manager adds the amount to your account after payment. You can see the balance after you sign in.

> **Am I charged for failed or refused images?**
>
> No. You pay only for images that are delivered. Check any request with the [cost endpoint](/errors-and-billing#check-what-a-request-cost).

> **Why does a request take up to a minute?**
>
> Image models render the full picture before the API answers. Nano Banana usually takes 10 to 60 seconds, GPT Image 2.5 about 30 to 70 seconds. Set your timeout to at least 180 seconds; 600 seconds is safest. See [Response time and timeouts](/images#response-time-and-timeouts).

> **How much does a video cost?**
>
> The per-second rate for the model and resolution times the duration. A 5-second Seedance 2.5 video at 720p costs \$1.525. The full price is held when the job starts and returned if it fails. See [Models and pricing](/models-and-pricing#video-models).

> **Can I get a video in one call?**
>
> Yes. `POST /v1/videos/generations` waits up to 600 seconds and returns the video. Set your client timeout to at least 620 seconds and send an `Idempotency-Key`. For production we recommend the async flow. See the [Video API](/videos).

> **Why did my video job fail?**
>
> The model can refuse a prompt, for example over copyrighted audio or other restricted content. The job ends with status `failed` and a full refund. Change the prompt and create a new job with a new `Idempotency-Key`. See [When the model refuses a prompt](/videos#when-the-model-refuses-a-prompt).

> **Can I edit an existing image?**
>
> Not yet. Image edits are coming soon. Today every model supports generations only.

> **Is there a webhook or async mode?**
>
> There are no webhooks. Images are synchronous: send requests in parallel for volume. Video runs as an async job you poll. See the [Video API](/videos).

> **How long is the image link valid?**
>
> Treat it as temporary. Download the file as soon as you get the response, or request `b64_json` and store the image yourself.

> **Can I limit how much one app spends?**
>
> Yes. Create a separate key per app in **Tokens** and set a limit on it.

> **Which size should I send?**
>
> Send the pixel size you need, such as `1536x1024` for a 3:2 image. Any size is accepted. Nano Banana models also take `aspect_ratio` plus `resolution`, for example `16:9` and `2k`. The longest side sets the price tier: under 1800 px (or `auto`) is 1K, 1800 px and up is 2K, 3000 px and up is 4K. The `quality` setting does not change the price.

> **How do I get help with a charge or an error?**
>
> Contact your Orvado manager and include the `X-Oneapi-Request-Id` of the request. It lets us find the exact call.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.