Privacy

PocketJev is designed for local, on-device inference.

• Camera captures and selected photos are processed by the model on the iPhone.
• The app does not contain an image-upload or telemetry path.
• The AI model is downloaded from Hugging Face the first time the user prepares it, so network access is required for that initial model download.
• After the model is cached, inference itself does not require a cloud inference API.

The current model is mlx-community/Qwen3-VL-2B-Instruct-4bit.

If you modify the app to add analytics, remote inference, logging, or uploads, this privacy description will no longer describe that modified build.