beatraDocs

Overview

Beatra gives agents production-grade multimodal creation — video, image, speech, and music. One install, one sign-in, one credit balance. These docs are written for two readers: developers integrating Beatra, and the agent that actually makes the calls.

How the three layers work together

All three paths land on the same async task: the same task fields, checking progress, callbacks, and billing. Prefer Skill + MCP; REST is for applications that do not go through an agent.

Choose your path

Capability map

Four modalities plus public social data. Each one is delivered as an async billed task with the same task fields, checking progress, and the same callback contract.

Capability pages are not tied to a specific model: the agent discovers models and prices live with beatra.models.list. To browse them directly, see the current model lineup.

Platform mechanics

MechanicWhat it answers
Async tasksHow a task moves from submit to a terminal state, how to poll, and what counts as failure.
Upload source mediaHow local images, audio, and video become reusable media IDs.
CallbacksHow a terminal state is pushed to your HTTPS URL, and how to verify the signature.
Billing & usageWhen credits are reserved, when they settle, and how failures refund.
Errors & retriesWhat the error envelope looks like, which errors retry, and how to branch.
Rate limitsConcurrency and rate ceilings, and how to back off after a 429.

Last updated

On this page

Markdown