MoneyPrinterTurbo

What it is

MoneyPrinterTurbo is an open source AI short video generator made by harry0703. You give it a topic or a keyword. It writes the script, matches footage, makes the voiceover and subtitles, adds background music and renders an HD video. You can also supply your own script, images and clips. It runs as a WebUI, an API, a command line, or through an AI agent that reads its Skill.

1. You give a topic or a keyword
2. A model writes the script and the footage search words
3. Footage is matched, and a voice reads the script
4. Subtitles and background music are added
5. The video renders as 9:16, 16:9 or 1:1

Who it suits

Creators and small teams who want many short vertical videos without opening an editor, and developers who want video generation they can call from a script. The README says you keep control of every stage, from the script to the editing.

Skip it if you want a hand-built timeline. MoneyPrinterTurbo assembles the video itself from a topic or your own script and clips, with a transition setting that config.example.toml lists as "None", "Shuffle", "FadeIn" and others. The repository’s separate editable project tool does not implement transitions or overlays in its first local renderer.

Where to use it

The README gives a route for each of these.

• Windows: a one-click package from GitHub Releases, with update.bat and start.bat
• macOS, Linux and Windows: the mpt package from pip, or the source with uv
• Docker: a Compose file that pulls a prebuilt image
• An AI agent that reads Skill documents and runs a terminal: macOS and Windows
• Google Colab: a notebook, with no local setup
• RecCloud: a free online generator that the README says is based on this project

The README recommends Windows 10 or later, macOS 11 or later, or a mainstream Linux distribution.

How to set it up

Get it saves a zip of Mindket’s setup notes in your downloads folder, with one file for each route the README documents. The README says Windows users should try the one-click package first, and that anyone with Python can install the pip package.

1. Install Python 3.11 or later and create a virtual environment.
2. Install the mpt package and start the WebUI. It opens in your browser.
3. Open the basic settings and add the keys for your model and your footage source.
4. Give it a topic and generate the video.

On macOS or Linux:

python3 -m venv .venv
source .venv/bin/activate
python -m pip install mpt
mpt webui

On Windows PowerShell, create and activate the environment with python -m venv .venv and .venv\Scripts\Activate.ps1, then run python -m pip install mpt and mpt webui.

Windows also has a shorter route. Download the .7z archive from the Assets section of the latest GitHub release, extract it, double-click update.bat, then double-click start.bat.

How to use it

Enter a topic in the WebUI. From a source checkout set up with uv, the README calls this the simplest complete generation command.

uv run python cli.py --video-subject "How AI is changing everyday life"

With the pip package, run mpt cli --help for the command-line options.

Or hand an AI agent the Skill. The README says the agent installs and configures MoneyPrinterTurbo, generates the video and returns the file path, and that it asks only for required keys that are not set yet.

Use this Skill: https://raw.githubusercontent.com/harry0703/MoneyPrinterTurbo/main/docs/skill/SKILL.md
Create a video with the topic "How AI is changing everyday life."

Example

You install the pip package and start the WebUI. In the basic settings you add a model key and a Pexels key. You enter the topic “How AI is changing everyday life”. The tool writes the script, fetches footage for its search words, reads it in a voice such as free Edge TTS, adds subtitles and music, and renders the video.

When an agent runs it, the Skill says a successful run prints lines of this shape:

MPT_RESULT
VIDEO_FILE=<absolute path>/final-1.mp4
TASK_DIR=<absolute path>/storage/tasks/<task_id>
LOG_FILE=<absolute path>/run-<task_id>.log
RESULT_FILE=<absolute path>/latest-result.json

VIDEO_FILE is the path of final-1.mp4, the finished video. The other lines name the task folder, the log file and the result file.

For newcomers and experts

If you are new. MoneyPrinterTurbo is a program that makes a short video for you. A language model writes the words, a voice reads them, and stock clips fill the screen. You usually add a model key and a free Pexels key. A ChatGPT login through the Codex provider needs no API key, and your own images and videos need no footage key. Start with the Windows package or the pip install, then add the keys in the basic settings. Edge TTS needs no key.

For experts. To run several tasks one after another from a source checkout, pass a UTF-8 JSON array or JSONL manifest. The README’s example is uv run python cli.py --batch-file ./tasks.json --stop-at video. A manifest holds up to 100 tasks and 1 MiB. All entries are checked before the first task starts, and the run goes on after one task fails. The command prints one JSON summary with total, succeeded, failed and tasks.

For subtitle style and voiceover options, an explicit CLI option beats the saved [ui] value in config.toml, which beats the built-in default. Background music, video count and paragraph count are not inherited from the WebUI. The subtitle mode edge uses TTS timestamps. The mode whisper runs faster-whisper locally and downloads a model of about 3 GB, or about 1.6 GB for large-v3-turbo, on first use.

Two ChatGPT sign-ins exist and they are separate. The Codex provider uses a device code, and the same login serves the Codex image source. The native OAuth provider needs an eligible account and a browser on the same computer, so it does not work from Docker or a remote machine. The repository also has a command-line tool for editable local video projects. It keeps revisions of prepared scenes and rebuilds only the stages a change touches.

Good to know

Check these before you publish anything.

• Package name: install mpt. The README says the MoneyPrinterTurbo package on PyPI is maintained by a third party
• Music: the README says the project includes some default music taken from YouTube videos, and to delete it if there are copyright issues
• Costs: Pexels and Pixabay footage is free. config.example.toml says WaveSpeed generation is billed per request and OFox generation per clip, and the Skill asks you to confirm before a paid video source runs
• ChatGPT logins: the Codex provider spends the connected subscription’s allowance, and each Codex image uses it too
• Keys: they live in config.toml, which the comments say not to commit
• Network: config.example.toml sets the API host to 0.0.0.0, which listens on all interfaces. It has an optional api_key, sent as the x-api-key header. docker-compose.release.yml publishes its ports on 127.0.0.1 only
• Data: the model you pick gets your topic and writes the script and search keywords. The voice service gets the script. Pexels, Pixabay and Coverr get keyword searches, and AI video sources such as Seedance work from script segments. The launch scripts turn off Streamlit usage statistics
• Windows package: the Source code archives on the release page have webui.bat but no start.bat or update.bat
• Removal: the README and the docs give no uninstall steps
• License: MIT. The LICENSE file reads “Copyright (c) 2024 Harry”