# AI media generation data API for AI agents

looot lists generation operations in its catalog. An agent generates an image from a prompt, a short video, music from lyrics or a prompt, or speech from text, and reads the result as a file or a link. Each call shows its price before it runs.

## Jobs an agent does

- Generate a hero image for a post from a text prompt.
- Make a short product teaser clip from a prompt.
- Turn a script into a voiceover.
- Produce a music bed from lyrics.
- Get speech with character timings to sync captions.

## Operations by job

- Video: A short video from a prompt. The run is asynchronous, so the agent reads the result when it finishes.
- Images: Images from a prompt with official image models such as FLUX, Ideogram and Recraft.
- Music: Music from lyrics or a prompt.
- Voice: Speech from text, as an audio file or with character timings for captions.

## How the providers differ

Replicate runs official models. Its image operation offers several of them, including FLUX and Ideogram, and the video and music operations each name their model in the description. ElevenLabs does speech in two forms, an audio file download, or the speech together with the timing of every character for captions and lip sync. Video runs are asynchronous, so the result arrives after the call returns.

## Prompt

Use looot to generate a product image of a ceramic mug on a white table with FLUX schnell, then a spoken caption for it with ElevenLabs. Show the price of each call before you run it. Setup is in skill.md.

## What can an agent generate through looot?

Images, short video, music and speech. The groups above list the operations that are live now.

## Which image models are available?

The image operation offers official Replicate models, among them FLUX schnell, FLUX dev, FLUX 1.1 pro, Ideogram v3 turbo and Recraft v3. You pick the model in the input.

## Why does a video run take longer?

Video generation is asynchronous. The call starts the run and the agent reads the result when it is ready.

## How do I get captions that match the speech?

Use the ElevenLabs operation that returns character timings with the speech. The other one returns an audio file you download.

## Do I need a Replicate or ElevenLabs account?

No. Your agent calls the operations with a looot key. You do not manage tokens for each service.

## Can I bring my own prompt and model inputs?

Yes. The operations take the model's own inputs, such as a prompt, and pass them on.

## How does my agent call a AI media generation operation?

It connects to looot once, searches the catalog for the job, reads the input fields and the price, then runs the operation. Setup takes one line from skill.md, and the same key works for every provider.

## How is a AI media generation call billed?

Per call, at the price shown on this page for that operation. The agent sees the price before the run, and the run is recorded with its cost in your ledger.

## Where do I find the full input and output fields?

Each operation has a reference page on docs.looot.ai with its input fields, an example call and the response. The For agents card on this page links the cheapest ones.

## Are the prices on this page current?

They are read from the live looot catalog every few minutes, so they change when a provider changes its price. When the catalog cannot be read, the page shows no numbers.

## Which AI media generation operation should I start with?

Start with the cheapest operation in the job you need, which is listed first in its group. Check that it returns the fields your agent reads, and switch to another provider if it does not.

## Related pages

- /platforms/youtube
- /platforms/tiktok
- /platforms/instagram
- /use-cases/data-enrichment

## Last reviewed

2026-10-04. The operation list, the counts and every price are read live from the catalog on the web page.

## For agents

Run any operation with POST /v1/runs and its endpointId. The full tool reference is on docs.looot.ai.

- **platform**: AI media generation
- **setup**: https://looot.ai/skill.md
- **tool reference**: https://docs.looot.ai/reference/tools