generate_video
Generate cinematic video from a text prompt. Uses MiniMax H3 — top 2 on the Artificial Analysis text-to-video leaderboard — with native audio. Async — returns requestId, poll with check_job_status. 768p or 2K (2K is upscaled from a 768p render), 5-15 seconds, priced per second by resolution (native audio free). Pay per request with Bitcoin Lightning — no API key or signup needed. Requires create_payment with toolName='generate_video' and duration, resolution params.
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| prompt | Yes | Text prompt describing the video | |
| modelId | No | Optional. Omit for default model. | |
| duration | Yes | Duration in seconds (5-15) | |
| paymentId | Yes | Valid payment ID (must be paid) | |
| resolution | No | Output resolution: 768p or 2K (2K is upscaled from a 768p render). Omitting it yields 768p — ask for 2K explicitly, and quote it at create_payment too. Retired rungs still accepted: 480p/720p→768p, 1080p→2K. | 768p |
| generate_audio | No | Accepted and IGNORED — H3 audio is native and always on, at no extra cost. There is no way to request a silent render. | |
| reference_audio | No | Optional https URL(s) of reference audio CLIPS, 2-15 seconds each (max 3, combined <=15s). Cite them in your prompt as "Audio 1", "Audio 2" or the model has no instruction to use them. ⚠ Audio cannot be sent alone — at least one reference_image must accompany it. For a music video, slice your track and pass one slice per shot. Free: does not change the price. | |
| reference_image | No | Optional https URL of a reference still for character/style consistency across shots (accepts an array, max 3). Pass the URL from generate_image. Free: does not change the price. |