A local-first short-video generator
MoneyPrinter takes a topic and produces a vertical short video with narration, captions, B-roll, and background music. It was one of the projects that popularized the AI-shorts idea, and its current form leans local-first: script generation and metadata run on local Ollama models, and a database-backed generation queue (API plus worker plus Postgres in Docker) makes processing restart-safe. At 13,530 stars as of 2026-06, it is the lighter, privacy-leaning option in this corner of GitHub.
The first thing to settle is which MoneyPrinter you mean, because the much larger MoneyPrinterTurbo is a separate project, not a fork.
What it does
You give it a subject, and it generates a script with a local LLM, narrates it, transcribes for captions, pulls stock footage to match, adds music, and composes a 1080x1920 short. The Ollama-first design means the language steps run on your machine without a cloud LLM bill, which is the main reason to prefer it over cloud-dependent alternatives. The queue architecture makes it more robust to restarts than a single-shot script.
Install
It uses Python with uv, needs Ollama running locally, and depends on FFmpeg and ImageMagick:
ollama serve
ollama pull llama3.1:8bYou then configure a .env with a stock-footage key (Pexels) and any other sources, and run the app and worker. ImageMagick is auto-detected from your PATH, with a manual override in .env if detection fails. Expect to install the system media dependencies before anything renders.
MoneyPrinter versus MoneyPrinterTurbo
This comparison is the one people actually search for. MoneyPrinterTurbo, despite the similar name, is an independent project created shortly after MoneyPrinter, not a fork of it. It is now far larger and more actively maintained, with a richer Web UI, many model backends, and easier setup paths. MoneyPrinter is the lighter, local-first original: simpler, Ollama-centric, and privacy-leaning, but with slower development. If you want the most features and active maintenance, Turbo is the answer; if you want a minimal, local-only setup, MoneyPrinter fits.
The risks worth naming
The honest part. Cost is modest because the LLM runs locally, though stock-footage APIs and any premium services have their own limits. The real risk is platform policy. YouTube and TikTok restrict mass automated and undisclosed AI-generated uploads, and bulk posting can lead to throttling, shadow-bans, or account bans. Stock footage from Pexels is generally free to use, but you are responsible for the rights of generated content and for disclosing AI generation where a platform requires it. Use this for personal, low-volume, human-reviewed output. Running it as an unattended content farm to flood a platform is against those platforms’ terms and is the use this page does not endorse.
Where it fits, and where it doesn’t
MoneyPrinter suits someone who wants a local, low-cost way to prototype short-video automation, who values keeping the LLM on their own machine, or who wants a smaller codebase to understand and modify.
It is a poor match if you want the most capable, actively developed tool, where MoneyPrinterTurbo is ahead, or if you are not prepared to install and troubleshoot media dependencies like FFmpeg and ImageMagick. And it is the wrong tool entirely for high-volume automated uploading, for the policy reasons above.
MoneyPrinter versus the alternatives
| MoneyPrinter | MoneyPrinterTurbo | ShortGPT | |
|---|---|---|---|
| Stars | 13,530 | 87,305 | 7,405 |
| Maintenance | maintenance mode | very active | stalled |
| Models | Ollama-first (local) | many backends | OpenAI, Gemini, EdgeTTS |
| Setup | local, uv, Ollama | one-click, Docker, Colab | Docker, fragile |
| Best for | local, privacy-first runs | most features today | reading the architecture |
Counts are from GitHub as of June 2026. MoneyPrinterTurbo is the larger, more active independent project, the default if you want capability and support. ShortGPT is an earlier framework whose maintenance has stalled. MoneyPrinter’s niche is the lightweight, local, Ollama-first run.
The star curve
The star curve rose early as the AI-shorts concept spread, then settled into a slower climb as the larger Turbo project absorbed much of the attention. The shape reflects an original that remains useful but is no longer the busiest project in its own niche.
What the issue tracker warns you about
The recurring issues are integration fragility. The stock-footage and music sources can fail with connection errors or dead links. Video composition through MoviePy can hang or produce delayed and missing captions. Free LLM workarounds and browser-automation pieces break as upstream services change. And the media dependency chain (FFmpeg, ImageMagick, MoviePy versions) is a common source of install failures. The local-first design avoids cloud LLM cost, but it does not avoid the brittleness of stitching media tools together.
Related
For the larger, more active sibling, see MoneyPrinterTurbo. For an earlier, now-stalled framework, see ShortGPT. For programmatic, code-defined video, see Remotion. For what is trending, see the daily digest and the weekly report.
FAQ
MoneyPrinter vs MoneyPrinterTurbo, which should I use? They are separate projects, not forks. MoneyPrinterTurbo is far larger, more actively maintained, and more capable, so it is the default. MoneyPrinter is the lighter, local-first, Ollama-centric original, better if you want a minimal local setup.
Does MoneyPrinter need an OpenAI key? No. It is Ollama-first, so script generation runs on a local model. You do still need a stock-footage key like Pexels and the media dependencies, but the language steps avoid a cloud LLM bill.
Is it against the rules to auto-upload these videos? Often, yes. YouTube and TikTok restrict mass automated and undisclosed AI uploads, which can trigger throttling or bans. Keep volume low, review output, and disclose AI generation where required.
What do I need to install for MoneyPrinter? Python with uv, a running Ollama with a pulled model, FFmpeg, and ImageMagick, plus a .env with a stock-footage key. Install the media dependencies first, since they are the common point of failure.
Is MoneyPrinter still maintained? It is in a slower maintenance mode, with updates less frequent than the actively developed MoneyPrinterTurbo. It works, but for ongoing development and features the larger project is the more active choice.