Create PDF and Word documents Turn your ideas into real documents in seconds: Create PDF and Word (.docx) files from text with a rich built-in editor: headings, lists, checklists, bold, code blocks and ready-to-use templates. Dictate instead of typing — voice to text, transcribed locally on your device. Add photos to your documents — images are embedded right into the PDF or Word file, always sized to fit the page and respect the margins. Edit anything, anytime Edit any document you create: change the text and regenerate the file instantly. Edit imported PDFs and Word files as text — the app extracts the content (even from scanned PDFs using on-device OCR) so you can rework it and export it again. Preview Word documents directly in the app, and use the full PDF toolkit: preview, compress, sign and share. Smart document library Powerful on-device search (BM25) that finds text inside your documents. Scan documents with your camera and keep them organized. Your documents stay safe even after app updates — no more lost or blank files. Edit photos with AI — 100% offline Generate and edit images with AI entirely on your device — no internet, no cloud, no uploads. Retouch photos, remove or replace parts of an image with simple prompts and masks, all powered by a local AI model that runs on your iPhone.

**What's New** - **Fixed context window selection**: Changing the context size (4K, 8K, etc.) in Settings no longer resets when loading a model. Your preference is now preserved. - **Improved stability**: Resolved a crash that occurred when sending messages with special characters (accents, ñ, etc.). The app now handles internal engine errors gracefully instead of closing unexpectedly. - **Refined document preview**: Removed the plain text display below the PDF preview for a cleaner, more spacious viewing experience. The PDF preview now expands to fill the full available height. - **Simplified document navigation**: Removed the "Summarize" tab from the Documents section for a more focused workflow. - **General reliability improvements**: Various low-level fixes to the inference engine for smoother performance.

Quick model switching in chat: Tap the model name pill in the chat header to instantly switch between any downloaded text or vision model Smarter model activation: Selecting an image in chat automatically activates your vision model. Going to Image Generation pauses your chat model and restores it when you return First-launch welcome: New users see a guided recommendation to download their first text model on first launch Vision model prompt: Tapping "Take Photo" or "Attach Image" shows a download prompt if no vision model is installed Audio model prompt: Enabling Text-to-Speech prompts you to download Whisper (STT) and the Piper TTS model for your system language if missing Image generation onboarding: Entering Image Generation with no models installed shows a prompt to download "Models Image", with a delete option once installed Voice language auto-detection: The voice language menu now shows the detected system language (e.g. "Auto (English)") when set to automatic Curated audio catalog: The audio model catalog now shows only Whisper Base and the Piper TTS model matching your device language for a cleaner experience Assistant/Agent mode redesign: Model and role selectors are now grouped above the Assistant/Agent mode toggle for a clearer layout Improved image model cards: Model cards in the image catalog now have a sleeker asymmetric layout with a dedicated delete button when installed Localization updates: Added translations for Polish, Chinese, French, Italian, and Arabic across multiple UI strings Bug fixes: Fixed a format specifier issue in Portuguese (Brazil) localization, renamed "Realvision" to "Models Image" in the image model catalog

- correction of errors in the chat - new models were added - the ui was updated - classification in the model catalog - widget fix