Ozyra FAQ
Can I trust that the AI is always correct?
No. Large language models can be wrong, outdated, or incomplete. Always double-check important answers. Enabling web search can help with fresh information but is not a guarantee of accuracy.
Why did I get an off-topic or odd response?
Models may hallucinate. Try rephrasing, providing more context, or regenerating. Verify outputs before relying on them.
Do you train models on my data?
We do not train our own models and have opted out of provider training where possible:
- OpenAI: Training disabled by default
- Anthropic: Training disabled by default
- Google: Training disabled on paid models
- OpenRouter: We avoid routing to providers that permit training
What AI models can I use?
Ozyra integrates models from leading providers such as Google (Gemini), Anthropic (Claude), OpenAI (GPT), DeepSeek, and more. Available models depend on your plan; premium unlocks the full catalog.
Where is my data stored?
By default, conversations live locally in your browser. If you sign in, data can sync to our cloud (Supabase) for multi-device access.
How do I delete my data or account?
Delete conversations or your account from settings. Synced data is removed from our systems subject to legal retention needs. For help, contact privacy@ozyra.com.
How does billing work?
Subscriptions are processed by Stripe and renew automatically. You can cancel anytime before renewal in settings; access continues through the current period. Refunds are handled case-by-case.
How is my data secured?
We use TLS in transit, provider-managed encryption at rest, and access controls. See our Security Policy for details.
How do I report a security issue?
Email security@ozyra.com with details and steps to reproduce. We appreciate responsible disclosure.
How do I get help or request features?
For support or feature ideas, visit our feedback page to submit requests or vote on existing suggestions.