Build with Vydra
Generate images, video, and speech through one API. Use the creative workspace yourself, integrate it into your application, or connect an agent.
https://vydra.ai/api/v1First API request
Create a key, select a model, submit a job, and retrieve its result.
Create and Studio
Use the workspace or execute a saved graph with input overrides.
Models and pricing
Nano Banana Pro, GPT Image 2, Seedance, and model-specific settings.
Media library
Upload references, organize media, and reuse generation outputs.
Speech and prompt refinement
Generate voiceovers or refine a prompt with the job API.
OpenClaw and MCP
Connect the same account to your agent tools.
How requests work
Create a key in your account and send it as a Bearer token from your server. For autonomous onboarding, use the bot registration flow in the authentication guide.
Submit a workflow to POST /jobs. The response includes a job ID and charged credits. Poll GET /jobs/{id} until completed, failed, or cancelled. Completed jobs expose media URLs or text in result. Image and speech requests can finish during the initial request; still retrieve the job to get its output.
Credits depend on the selected model and settings. Reading the catalog or polling status does not create a new paid generation. Each variation and each submitted Studio generation is a separate job. See billing and credits for the plan conversion and refund behavior.
Reference and discovery
Capabilities describes available API operations. GET /creator/models returns the shared Create/Studio catalog with supported controls and default estimates. The legacy /models registry is a separate discovery surface; use the request contract for the endpoint you call.
Download the OpenAPI specification or use the agent guide. Studio and library endpoints require the creator platform rollout on your target deployment; a local build does not establish hosted availability.