OpenPocket

OpenPocket brings private AI conversations to your iPhone and iPad. Explore your ideas while keeping your conversations local and private on your device.

A little intelligence. All yours.

OpenPocket brings private AI conversations to your iPhone and iPad. Use Apple’s on-device model where available, or download a compatible model and explore your ideas without sending your conversation to a cloud AI service.

FIND YOUR MODEL Explore models with recognizable icons, colorful capability tags and guidance based on your device’s memory and storage. Browse downloadable 4-bit models, manage your downloads and switch between available models.

APPLE AI, BUILT IN Use Apple’s AFM 3 Core for conversations, document tasks and image questions on supported devices. OpenPocket checks availability and explains when the model isn’t ready. Apple manages the model files, so there’s no separate model download inside OpenPocket.

Requires iOS or iPadOS 27 or later, Apple Intelligence enabled and AFM 3 Core available on your device. Availability depends on your device, region and language.

KEEP YOUR TRAIN OF THOUGHT Ask questions, organise your thoughts and find the right words. Save conversations on your device, search them, rename them and pick up where you left off.

BRING YOUR OWN WORDS AND IMAGES Import text, Markdown or text-based PDFs. Review and edit the extracted text before adding it to a conversation. Choose a photo from your library or take one with the camera, then ask a compatible vision model about it. Images are resized for on-device processing. Input limits depend on the model, and scanned PDFs are not supported.

SPEAK, REVIEW, SEND Tap the microphone to dictate a message, stop when you’re finished, then review and edit the text before sending. Dictation uses on-device speech recognition where supported by your device and language.

LISTEN YOUR WAY Have responses read aloud using available system voices. Adjust the speaking rate, pause playback or stop whenever you like.

TURN IDEAS INTO IMAGES Experiment with local image generation using Stable Diffusion 1.5. Describe an image, adjust generation settings and save or share your creations. This experimental feature requires a separate model download and a compatible device. Generation speed and results vary.

LOCAL BY DESIGN No account, analytics or cloud conversation sync. Model setup needs an internet connection and enough free storage. Once a compatible model is ready, you can chat offline. You choose which conversations, generated images and downloaded models to keep.

AI responses can be inaccurate or inappropriate. Check important information and use the flag beneath a response to prepare an optional report. Nothing is sent automatically.

Model availability, speed, memory use and battery use vary by device. Larger models, longer conversations and image generation need more resources.