Batch-generate Gemini Omni Flash audio-native video through the Google Flow API by useapi.net — synced dialogue, reference images, and video-to-video edits from a list of prompts.
📖 Full walkthrough: Generate audio-native AI video with Gemini Omni Flash
omni-flash.mjs reads prompts from prompts.json, uploads any reference images or source videos, submits each job to POST /videos in async mode with model: "omni-flash", polls GET /jobs/{jobId}, and downloads every finished MP4.
- Node.js v21 or newer (no dependencies to install — uses built-in
fetch) - A useapi.net API token
- A connected Google Flow account email (Omni Flash runs on Plus, Pro, and Ultra plans)
node ./omni-flash.mjs <API_TOKEN> <EMAIL> [PROMPTS_FILE]PROMPTS_FILE defaults to prompts.json. The script looks the account up by email and checks its health field before submitting.
prompts.json is an array of prompt objects — prompt is the only required field; everything else falls back to the API defaults (model omni-flash, landscape, 8 seconds; durations 4 / 6 / 8 / 10 s).
- Spoken dialogue: set
referenceAudio_1to a preset voice name (e.g.Charon,Kore) or aPOST /voicesuser-voice id. - Reference-to-video: use
referenceImage_1…referenceImage_7(local file paths, uploaded for you). - Video-to-video edit: set
referenceVideo_1to a local MP4 plusstartFrameIndex_1/endFrameIndex_1for the trim window (output max 10 s).
Every parameter is documented on POST /videos. Local image/video paths in prompts.json (e.g. ./subject.jpeg, ./source.mp4) are inputs you supply — they are not included in this repo.