Because AI-generated content varies each time — different script phrasing, different footage clips selected — running a batch lets you compare several versions and choose the one you like most.
How it works
Set video_count to the number of variants you want. MoneyPrinterTurbo runs the full pipeline once per variant, with each video receiving a different random selection of footage clips (when using video_concat_mode = "random").
Output files are named sequentially: final-1.mp4, final-2.mp4, and so on.
Using batch generation
In the Web UI, look for the Number of videos (or video_count) slider or input field and set it to the desired number before clicking Generate.
Pass video_count in the request body when calling POST /api/v1/videos:curl -X POST http://localhost:8080/api/v1/videos \
-H "Content-Type: application/json" \
-d '{
"video_subject": "benefits of morning exercise",
"video_count": 3,
"video_aspect": "9:16"
}'
task_id. Poll GET /api/v1/tasks/{task_id} until completion — the response will include all generated video URLs:{
"status": 200,
"data": {
"state": 1,
"progress": 100,
"videos": [
"http://localhost:8080/tasks/abc123/final-1.mp4",
"http://localhost:8080/tasks/abc123/final-2.mp4",
"http://localhost:8080/tasks/abc123/final-3.mp4"
]
}
}
Concurrency and server load
The max_concurrent_tasks setting in config.toml controls how many video generation tasks run in parallel across all users (default: 5). Each video in a batch counts as one task step, so generating large batches on a low-powered machine may be slow.
[app]
max_concurrent_tasks = 5
Start with video_count = 2 or 3 to find the right balance between variety and generation time. For social media content, 2–3 variants is usually enough to pick a winner.
Downloading batch results
You can download any video from the batch via its URL in the task response, or use the streaming endpoint to preview in the browser:
# Download a specific variant
curl -O http://localhost:8080/api/v1/download/abc123/final-2.mp4