FAQ
Questions, answered.
Straight answers about the beta. If something here is wrong or missing, tell us on GitHub.
Is CueTwin free?
Yes, CueTwin is free during the public beta, with no trial timer. The free tier exports up to 1080p and includes instant voice cloning; higher resolutions and professional voice training will need a paid Creator licence. Paid licences come later through Lemon Squeezy; pricing has not been decided and nothing is on sale yet.
Can I download it today?
Yes. Grab the latest beta from the Download page or the GitHub releases page.
Which platforms are supported?
The beta targets Windows 10 and 11, 64-bit.
How big is the download?
The installer itself is kept small enough for GitHub Releases. The AI models are not bundled: on first launch CueTwin downloads its local AI models (roughly 8 GB), so you need an internet connection and at least ~8 GB of free disk space, plus space for your projects. After that, the assistant runs locally. Exact sizes will be confirmed with the first build.
Does it work offline?
Editing, the built-in assistant and voice generation run on your computer once the app and its models are installed. You need the internet for the first-launch AI model download (roughly 8 GB), for voice setup (Python packages and speech models), for optional cloud AI backends, and for licence checks once paid licences arrive.
Do I need an API key or AI credits?
No. The built-in CueTwin assistant runs locally with no API key and no usage credits. Optional backends (Claude, Cursor CLI, Ollama, OpenAI-compatible services) use your own account or installation, and any charges are billed by those providers.
Why does voice cloning need Python?
The local voice runtime runs on Python 3.11 (64-bit). CueTwin finds your Python install, creates its own environments, installs the dependencies (NVIDIA GPU or CPU builds) and downloads the Qwen speech and Whisper models. Install Python 3.11 from python.org and tick "Add to PATH" first.
Do my voice recordings leave my PC?
Voice recordings and training stay on your computer. Only clone your own voice, or a voice you have permission to use.
Is the built-in assistant a custom-trained model?
No. It uses CueTwin's editing tools and instructions with local Qwen base models (Qwen3.5 9B or 4B, depending on your GPU). It has not been custom fine-tuned.
Does CueTwin ever delete my source files?
No. Removing media from a project takes it out of the project and off the timeline, but the file on disk is never touched. Imported .mkv files are remuxed into an mp4 preview proxy; the original is not modified.
How fast is it?
Speed depends on your hardware and what else is using your GPU. CPU-only operation works but is slower, and real-time speech generation is not promised. Measured requirements will follow clean-PC testing.
How do I report a bug?
Open an issue on GitHub. Include the app version, your Windows version, CPU, GPU and RAM, numbered steps to reproduce, what you expected versus what happened, and any error text. Please do not post licence keys, API keys or private recordings.
System requirements
What you’ll need
| Component | Recommended (beta target) |
|---|---|
| Operating system | Windows 10 or 11, 64-bit |
| Memory | 16 GB RAM |
| Graphics | A compatible GPU is recommended. CPU-only operation is supported, but slower; real-time speed is not promised. |
| First run | An internet connection. On first launch CueTwin downloads its local AI models (roughly 8 GB); after that the built-in assistant runs locally. |
| Voice cloning | Python 3.11 (64-bit) from python.org, plus an internet connection to install the voice runtime and download the speech models |
| Disk space | At least ~8 GB free for the AI models, plus space for your projects (and the voice models, if you set up voice cloning). Exact figure confirmed with the first build. |
These are the beta’s target specs, not measured minimums. Professional voice training can need more memory and compute than generating speech.