Privacy Policy

Overview

LocalGemmaChat is a local-first LLM chat application. Your privacy is our top priority. This policy outlines how we handle information and the technical nature of our application.

2. Network Access & Data Flow

The app only makes network requests for the following specific purposes:

• Model Downloads: When you select a model, it is downloaded from external repositories (e.g., Hugsting Face/hf-mirror.com). This is a one-time download per model.
• Model Metadata: Minimal requests to fetch basic model info (size, architecture) before the download process begins.

Note: Once models are downloaded, you can disable network access entirely. The app is designed to function fully offline.

3. Local Storage

All data remains within your device's private application container:

• Conversation History: Stored locally in the app's private storage.
• Model Weights: Stored in the system cache directory.
• Data Control: You can delete all data at any time via the "Clear Chat" feature or by uninstalling the app.

4. Third-Party Components

This app utilizes the following technologies for local execution:

• llama.cpp (C++ inference engine)
• MLX (Apple's machine learning framework)
• FoundationModels (Apple's on-device LLM framework)

These components run locally on your device. No data is sent to their maintainers. When downloading models, your IP address may be visible to the host provider (e.g., Hugging Face) solely for the purpose of the download session. We do not control or store this connection data.

5. Your Rights & Compliance

Your Rights: Since we do not collect or transmit personal data, there is no data held on our servers to access, correct, or delete. Your data belongs solely to you and stays on your device.

Children's Privacy: This app is intended for general audiences and does share personal information with third parties.