Capabilities overview
Every Beatra capability in one matrix: modality, MCP tool, REST endpoint, execution mode.
One row per capability. Agents pick a tool here, then open the capability page
for arguments, task status, and recovery. Models, controls, and prices are
discovered live with beatra.models.list; to browse what's available right
now, see the current model lineup.
Images
| Capability | Input → Output | MCP tool | REST endpoint |
|---|---|---|---|
| Image understanding | 1–8 images + text → text | beatra.images.understand | POST /v1/images/responses |
| Text to image | text → image | beatra.images.generate | POST /v1/images/generations |
| Image to image | image + text → image | beatra.images.transform | POST /v1/images/image-to-image |
| Image edit | image + text → image | beatra.images.edit | POST /v1/images/edits |
Videos
| Capability | Input → Output | MCP tool | REST endpoint |
|---|---|---|---|
| Video understanding | one video + text → text | beatra.videos.understand | POST /v1/videos/responses |
| Video prompt enhancement | text + optional references → enhanced text | beatra.videos.enhance_prompt | POST /v1/videos/prompts/enhance |
| Text to video | text → video | beatra.videos.generate | POST /v1/videos/text-to-video |
| Image to video | image + text → video | beatra.videos.animate | POST /v1/videos/image-to-video |
| Reference to video | references + text → video | beatra.videos.generate_from_references | POST /v1/videos/reference-to-video |
| Frames to video | required last frame + optional first frame + text → video | beatra.videos.interpolate | POST /v1/videos/frames-to-video |
| Video edit | video + references + text → video | beatra.videos.edit | POST /v1/videos/edit |
| Video extend | video + text → video | beatra.videos.extend | POST /v1/videos/extend |
Speech & voices
| Capability | Input → Output | MCP tool | REST endpoint |
|---|---|---|---|
| Text to speech | text + voice → audio | beatra.speech.synthesize | POST /v1/audio/speech |
| Voice clone | audio sample → voice | beatra.voices.clone | POST /v1/voices |
| Voices | — (management) | beatra.voices.list | GET /v1/voices |
Music
| Capability | Input → Output | MCP tool | REST endpoint |
|---|---|---|---|
| Text to music | prompt / lyrics → audio | beatra.music.generate | POST /v1/music/text-to-music |
| Reference audio to music | audio + text → audio | beatra.music.generate | POST /v1/music/reference-audio-to-music |
Public social data
| Capability | Input → Output | MCP tool | REST endpoint |
|---|---|---|---|
| Public social data | public social query → JSON (text, metrics, media URLs) | beatra.social.tools.search / get / execute | MCP only |
Social reads reuse the existing OAuth MCP connection. Each execute is one
prepaid task priced at 6, 12, 30, or 60 credits. There is no public Social REST
API.
Shared contract
Every create or understanding call above is billable and returns an asynchronous
task: 202 + a Task envelope, polled with beatra.tasks.get (or
GET /v1/tasks/{task_id}) until terminal. Discovery and management reads
(voices, models, tasks) are synchronous. The full lifecycle — envelope schema, cancellation,
refunds — is defined once in Async tasks and
applies to every row.