harry0703/MoneyPrinterTurbo

What it does

MoneyPrinterTurbo turns a topic or keyword into a finished short video. The pipeline writes a script with a large language model, fetches or generates footage, synthesizes voice-over, adds subtitles and background music, and renders the final clip with FFmpeg. It targets people making vertical, horizontal, or square social videos for TikTok, Instagram Reels, and YouTube Shorts, and it exposes the workflow through a WebUI, a REST API, a CLI, and an agent skill so it can also be embedded in automation (README overview).

Quick start

Python 3.11 is required; the documented package route does not require cloning the repo. Create and activate a virtual environment, then install the official PyPI package mpt and launch the web UI (pip install steps):

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

On Windows PowerShell, activate with .venv\Scripts\Activate.ps1 instead. The first launch creates config.toml; enter API keys for the online LLM, TTS, or stock-footage providers you plan to use in the WebUI settings. Generated files land under storage/tasks in the platform data directory (for example ~/Library/Application Support/MoneyPrinterTurbo on macOS, %LOCALAPPDATA%\MoneyPrinterTurbo on Windows), unless MPT_HOME overrides it. Alternatives include a Windows all-in-one archive from releases, a Colab notebook, Docker Compose against the prebuilt image, or source deployment with uv sync --frozen. Without a browser, a one-shot CLI run looks like uv run python cli.py --video-subject "..." from the cloned directory.

Requirements and limits

Recommended systems are Windows 10, macOS 11+, or mainstream Linux; minimum 4 CPU cores and 4 GB RAM, with 8 GB or more recommended. A GPU is optional but helps with local Whisper transcription and batch generation. Edge TTS works without an API key, but most LLM and TTS providers, Pexels/Pixabay stock sources, and text-to-video services need their own credentials and may incur usage costs. Subtitle accuracy depends on the chosen provider: edge uses TTS timestamps, while whisper downloads a Faster-Whisper model (roughly 3 GB for large-v3, about 1.6 GB for large-v3-turbo) and benefits from a GPU. Windows paths should avoid Chinese characters, special characters, and spaces. Bundled background music is sourced from YouTube and the README asks users to remove it if there is a copyright issue. The project is MIT-licensed.

HotRepo's take

The repository has accumulated a large following: by the November 4, 2026 daily record it had 138,537 stars and 22,120 forks, with 615 stars added that day and a rank of 4 on the Python daily list. Earlier all-language daily records in August 2026 show gains in the thousands per day, while the September 30 and October 1 snapshots show 338 and 431, so the growth pace has varied substantially across the recorded days. The README spends a long section on sponsor placements and model-service promotions, which is worth knowing when evaluating which provider suggestions are editorial versus paid; the core workflow, installation steps, and configuration are documented independently of those placements.