> This page is for version v2 (default).
> For other versions, use one of these documentation indexes:
> - v2 (default): https://developer.ideogram.ai/v2/llms.txt
> - v1: https://developer.ideogram.ai/v1/llms.txt

> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://developer.ideogram.ai/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://developer.ideogram.ai/_mcp/server.

> The Ideogram API has no cancel operation. How to stop waiting for a request safely, handle a late result, and avoid paying for work you no longer need.

The Ideogram API doesn't support cancelling a request. Once a request is accepted, it runs to completion and is billed when it completes, even if you stop waiting for it. A request that is rejected before it runs, such as a `429`, is not billed.

## Stop waiting safely

* **Set a deadline in your client.** For an asynchronous request, poll [`GET /v2/generations/{generation_id}`](/v2/api-reference/generations/get-generation) until the result arrives or your own deadline passes. Keep the `generation_id` either way.
* **Expect a late result.** A generation you stopped waiting for can still finish. If you use [webhooks](/v2/ideogram-api/webhooks), verify the delivery, then ignore it or store it for later, based on the `generation_id` you recorded.
* **Don't resubmit to "restart".** Sending the request again starts a second generation, billed separately. If you still want the result, poll the first `generation_id`.

## Avoid paying for work you don't need

* **Quote first.** [`dry_run=true`](/v2/ideogram-api/price-quotes) returns a request's price without running it.
* **Send only what you will use.** Ask for the number of images you need and the quality you need.
* **Keep concurrency bounded.** A queue that sends requests as capacity frees up is easier to stop than a burst of requests sent at once. See [Rate limits](/v2/ideogram-api/rate-limits).