Privacy Policy for Chat LLM

Last updated: 11 September 2026

Chat LLM is developed and maintained by David Picknell ("we", "us", "our"). This Privacy Policy explains how the app handles your data and what information may be transmitted to external services when you use its features.

Useful links:
- Support Page
- Terms of Use

1. On-Device Local AI Model

Chat LLM offers an optional on-device AI model that runs entirely on your device. The model is not bundled with the app: it is downloaded on demand from Hugging Face the first time you tap the Local LLM card in the model picker. After this initial download, the model runs locally and works offline.

• The download itself contacts Hugging Face to fetch the model files; no chat content is sent during this step
• The local model generates its replies entirely on your device
• No chat content is ever sent to any AI provider or to us when using the local model
• The app can function offline once the model is on the device
• No account or API key is required
• Prompts and responses are kept on your device; if you use iCloud, your conversations also back up to your own private iCloud (see Section 8)
• You can delete the downloaded model from the model picker at any time to reclaim storage

We do not have access to any conversations you have with the local model.

2. Free Cloud AI Access (Groq)

Chat LLM provides free cloud AI access without requiring an API key. When using free cloud AI:

• Your text prompts and conversation content are sent to our backend server
• Requests are forwarded to Groq to generate AI responses
• Responses are returned to your device
• No account or API key is required

We do not store your prompts or responses. Messages are processed in real time and discarded after delivery. Free cloud access is text-only and does not support image or document uploads.

3. Reader and Text-to-Speech

Chat LLM's Reader is an audiobook-style player for PDFs, EPUBs, TXT files, pasted text, and webpages or documents shared into the app. Reader is also where any AI reply is read aloud (you share a message into Reader). Reader speaks text using a text-to-speech provider that you choose, per document, using your own API key for that provider:

• OpenRouter (Kokoro voice model) — requires your own OpenRouter API key
• MiniMax — requires your own MiniMax API key
• ElevenLabs — requires your own ElevenLabs API key
• Fish Audio — requires your own Fish Audio API key
• xAI — requires your own xAI API key, or sign in with your xAI account to use your xAI subscription

• The text being synthesised is transmitted directly from your device to the selected provider using your own key or sign-in; document text is processed locally and only the text needed for the current and upcoming sentences is sent
• Generated audio is downloaded to your device and cached locally so repeat playback does not re-incur a generation cost
• You can export a whole document to a single audio file that is saved on your device; exporting sends the document text to your selected provider the same way playback does, and the resulting audio file is stored only on your device (it is not uploaded to us)
• Generation costs are billed to your own provider account at that provider's standard rate
• Your provider API keys are stored only on your device
• When you import a webpage, the page is cleaned on your device by default; if you optionally add your own Groq API key, the webpage text may be sent to Groq to be tidied up for clearer narration

We do not store message text, document text, or audio on our servers.

4. AI Image and Video Generation

Chat LLM can generate images and videos into the chat thread using third-party providers you choose. When you generate media:

• Your prompt and any reference images you attach are transmitted directly from your device to the selected provider — OpenRouter (with your own OpenRouter API key) or xAI (with your own xAI API key or your xAI sign-in)
• OpenRouter may route the request to different model providers depending on the model selected
• The finished image or video is downloaded to your device and stored in the conversation it was requested from
• Generation costs are billed by the provider to your own account at that provider's rates

We do not store media prompts, reference images, or generated images or videos on our servers.

5. Grok Text Chat (with xAI Sign-In)

If you sign in with your xAI account, Grok text models become available in the model picker. When you chat with a Grok model:

• Your messages and conversation content are transmitted directly from your device to xAI to generate AI responses, using your xAI subscription
• Sign-in uses xAI's device-code flow; the resulting sign-in token is stored only on your device (in the iOS Keychain) and can be removed by signing out in the app
• If you have also entered an xAI API key, the key is used instead of the sign-in

We do not store your messages or responses on our servers.

6. LM Studio — Your Own Computer

Chat LLM can connect to LM Studio running on your own computer and use models hosted there. When you use this feature:

• Your chat content is sent over your own private network connection (for example, a Tailscale connection) directly to your own computer
• It is not sent to us, and it is not sent to any third-party cloud AI provider
• The connection endpoint and any access token you configure are stored only on your device
• If you configure MCP tool server names (from your LM Studio configuration), the model can invoke those tools; any external requests the tools make (for example, a web search) originate from your own computer, not from your phone

7. Voice Chat

Chat LLM supports real-time voice conversations through two optional backends. Gemini requires your own Gemini API key; Grok requires your own xAI API key, or you can sign in with your xAI account to use your xAI subscription instead of a key.

• Google Gemini: your spoken input, conversation content, and any images or photos you attach during a voice session are transmitted to Google to generate AI responses. Gemini may automatically perform Google Search queries to answer questions about current events, weather, or other live information. Your Gemini API key is stored only on your device.
• xAI Grok: your spoken input and conversation content are transmitted to xAI to generate AI responses. Your xAI API key is stored only on your device. If you sign in with your xAI account instead, the sign-in uses xAI's device-code flow and the resulting token is stored only on your device (in the iOS Keychain); you can remove it by signing out in the app.
• Audio is processed in real time for speech recognition and response generation
• Conversation transcripts may be stored locally on your device

We do not store voice data or transcripts on our servers.

8. iCloud Sync (Apple iCloud)

Chat LLM backs up your settings, system prompts, conversations, and imported Reader documents to your own iCloud account so they restore across your devices and after reinstalling the app.

• What is synced: app settings, generation settings, reading preferences, your selected voices, all of your system prompts (including the per-provider prompts and the secondary prompts used by the voice providers), your conversations, and your imported Reader documents (your Reader library)
• This uses Apple's iCloud, tied to your Apple ID; the data is stored in your own private iCloud, not on our servers, and we have no access to it
• If you are signed out of iCloud or it is unavailable, the app still works and your data is kept locally on your device

9. Bring Your Own API Key (Advanced Option)

Chat LLM allows you to connect your own API keys to access additional third-party AI services. If you choose to use this feature:

• Content is transmitted directly from your device to the selected provider
• Prompts, images, and documents are handled by that provider
• API keys are stored only on your device

These providers may include OpenRouter (text chat, web search, and image and video generation), Google Gemini (voice chat), xAI (voice chat, and — with a key or sign-in — Grok text chat, media generation, and Reader text-to-speech), OpenRouter (Kokoro), MiniMax, ElevenLabs, and Fish Audio (Reader text-to-speech), plus an optional Groq key for webpage import cleanup and a connection to your own LM Studio server. If you use OpenRouter, OpenRouter may route requests to different model providers depending on the model selected. When OpenRouter web search is enabled, search queries are processed by OpenRouter as part of the request; the query is augmented with your device's country and timezone so search results are region-relevant, and no other personal information is included.

We do not receive, store, or log content sent using your own API keys.

10. How Your Data Is Used

Chat LLM does not create user accounts, track usage, or collect analytics. We do not sell or share user data with advertisers.

The app does not do the following:

• Store prompts, images, documents, voice data, or generated audio or video on our servers
• Store API keys or sign-in tokens on our servers
• Track users across apps or websites
• Include advertising or analytics SDKs

11. Third-Party Data Handling

While we do not retain your content, third-party AI providers and services may process or retain data according to their own privacy policies and terms:

• Groq: https://groq.com/privacy-policy/
• Hugging Face (local model download only): https://huggingface.co/privacy
• Google (Gemini): https://policies.google.com/privacy
• xAI: https://x.ai/legal/privacy-policy/ and https://x.ai/legal/terms-of-service/
• OpenRouter: https://openrouter.ai/privacy
• MiniMax: https://www.minimax.io/privacy-policy
• ElevenLabs: https://elevenlabs.io/privacy
• Fish Audio: https://fish.audio/privacy/
• Apple iCloud: https://www.apple.com/legal/privacy/

12. Security

All network communication uses secure HTTPS connections. Local data remains on your device unless transmitted to a selected AI provider.

13. Your Rights

Because we do not store personal data on our servers, there is no user data held by us to access, export, or delete.

14. Contact

Email: ferrraridave@icloud.com

15. Changes to This Policy

We may update this Privacy Policy from time to time. Continued use of the app indicates acceptance of the updated policy.