generate_video(prompt, model="seedance_2_5", ...) and generate_image(prompt, model="nano_banana_2_lite", ...) wait for the run and return the file URL; submit, status and credits are there for finer control. All of them read EMAKI_API_KEY (create a key).
Python
seedance_2_5); other ids are on the models page. Extra keyword arguments go into parameters as they are. Set EMAKI_API_KEY or pass api_key=.
From an agent instead of code: the MCP server. Writing your own client: the HTTP API is two calls. For image to video, pass a first-frame URL as image.