Reference

Privacy architecture, frequently asked questions, troubleshooting, and support.

Privacy architecture

OpenChat has no backend server. The app on your iPhone is the entire system: there are no OpenChat-operated servers that receive your data.

What stays on your device

• Chat history, rules, memory, and skills (SwiftData)
• API keys and search keys (iOS Keychain)
• Local notifications and Live Activities

What leaves your device

Only requests you initiate, sent directly to providers you configured:

• Chat messages and context to your chosen AI provider
• Search queries to your chosen search provider (when web search is enabled)
• Images or documents you attach (to vision-capable models)

OpenChat includes no analytics SDKs, advertising libraries, or tracking. The app is open source (MIT): you can verify this in the source code.

Read the full privacy policy for details on third-party data handling, photos, camera, microphone, and notifications.

FAQ

Does OpenChat cost money?

The app is free. You pay your AI and search providers directly for API usage at their standard rates.

Do I need an OpenChat account?

No. OpenChat is bring-your-own-key. You only need API keys from the providers you want to use.

Which providers are supported?

OpenAI, Anthropic, Google Gemini, OpenRouter, Mistral AI, DeepSeek, Alibaba Cloud (Qwen), Moonshot AI (Kimi), Z.ai, 01.AI, Cheaper Inference, and any OpenAI-compatible custom endpoint. OpenRouter is the fastest way to reach DeepSeek, Qwen, Mistral, and Llama on one key. See Getting started → API keys for key creation links.

Where is my data stored?

On your iPhone. Chats use SwiftData. Keys use the Keychain. Nothing syncs to OpenChat servers.

Can I use multiple providers at once?

Yes. Connect as many providers as you want. Each conversation uses one model from one provider, but you can switch models anytime.

Does OpenChat train on my chats?

OpenChat does not collect or process your data. Whether a third-party provider retains or trains on API data depends on that provider's policies. Check their documentation.

Is web search required?

No. Web search is entirely optional and requires a separate search API key.

Troubleshooting

Model list is empty or still loading

OpenChat fetches model catalogs when you save a provider key. If the list is empty, check your internet connection and verify the API key is valid. For custom endpoints, model IDs must be entered manually.

API errors or authentication failures

• Confirm the key is correct and has not expired
• Check the provider's status page for outages
• For custom endpoints, verify the base URL and that the server is reachable from your device
• Re-enter the key in Settings → Providers → [provider]

Images not sending

Only vision-capable models accept images. Switch to a model with the vision capability badge, or remove attachments before sending.

Web search button is disabled

Add a search API key in Settings → Web Search and enable web search. Then pick a provider from the web search button in the composer.

First launch hangs after connecting OpenRouter

OpenChat starts a chat immediately with OpenRouter Auto while the full catalog loads in the background. If you see a delay, wait a few seconds for the catalog to cache, or pick a specific model from the picker.

Background reply did not finish

Ensure OpenChat has notification permissions. Very long generations may time out depending on the provider. Return to the chat to check status.

Source code and support

• Source code: github.com/robert-garay/openchat
• Issues and bug reports: github.com/robert-garay/openchat/issues
• Email: robert@genion.ai
• Privacy policy: robert-garay.github.io/openchat/privacy-policy.html