FAQ
DocsFAQ
HelpFAQ
Is Opera free?
Opera is open source under the GNU AGPL v3. You can build it yourself and run it on your own hardware. Using a hosted model provider may cost money depending on that provider.
Do I need an account?
No. Opera works without a sign-up. Sync is optional and can use a Sampler Hub, Nextcloud or Google Drive.
Does it work offline?
Yes, with a downloaded on-device model or a local server on your network. A cloud provider naturally needs a connection.
Where is my data stored?
Conversations and attachments live in the app’s private SQLite storage. On-device models are local files. Nothing is uploaded unless you enable sync, a backup, sharing or a cloud provider.
Which models can I use?
- Gemini Nano through Android AICore, or the browser’s built-in model.
- On-device
.litertlmmodels via LiteRT-LM. - Ollama.
- Any OpenAI-compatible endpoint, including LM Studio and self-hosted gateways.
See Models.
What is the difference between Opera and Sampler Hub?
Opera is the assistant app. Sampler Hub is a self-hosted server for encrypted sync, profiles and shared local models. You can use Opera without a hub.
Can I use Opera on several devices?
Yes. Pair them with a Sampler Hub, or move data with an encrypted cloud backup or a JSON export. See Sync and sharing.
Is sync end-to-end encrypted?
Sampler Hub sync is sealed end to end with the profile key. Nextcloud and Google Drive backups are encrypted with a PIN before upload. A JSON export is not encrypted by default.
What happens if I forget my Sampler Hub password?
If no paired device remains, the profile cannot be decrypted. There is no account recovery. Keep a device paired or maintain a backup.
Which platforms are supported?
Opera runs on Android, the web, and as Opera Desktop on Windows (x64 and ARM64), macOS (Apple silicon) and Linux (x86_64 and arm64). Sampler Hub runs on Windows, macOS (Apple silicon) and Linux (x86_64 and arm64). Check the releases page for the latest builds.
Which languages does the interface support?
English and French.
Can I turn tools off?
Yes. Every tool has its own toggle, and permissions are requested only when a feature needs them. A disabled tool is not offered to the model.
How do I report a bug or contribute?
Open an issue or a pull request on GitHub. For Sampler Hub issues, use the Sampler Hub repository.
Why is the first on-device reply slow?
The model has to load and warm up. Later replies in the same session are faster. A smaller model or a device with more memory will also help.