- Updated the UI in the online AI services - Added voice input and support for multiple languages - Fixed critical issues on Mac OS version to allow users to manage downloaded models

* The RAG strategy has been updated behind LLM, Better AI response qualities when asking for documentation. * The AI chat has been updated, Better conversation in local LLM. * Online AI summary support has been added * fix some issues * support speaker diarization in audio transcription

Fix UI issues and enhance performance.

* Enhance the prompts for text enhancement. * Allow macOS users to use Command Key for activating dictation.

* add voice to text feature for text input in AI chat view and Agent view * refine the localization * enhance UX

* fix issues in AI reprocessing view * refine localization * enhance UI

* Fix issues in the text enhancement module. * Use the Gemma model as the default bundle model to reduce package size.

* add agent mode * fix UI issues

* Improve UI performance to avoid lagging issues and enhance the overall user experience. * Fix issues in the text editor to ensure a smooth and efficient workflow.

• The text enhancement engine has been optimized to provide better performance, resulting in faster and more accurate text processing. • The shortcut mode has been improved to ensure stability and consistency, making it easier to use and navigate. • The real-time translation chat has been enhanced to provide a more stable and seamless translation experience. • The user interface in post-processing views has been improved to provide a more intuitive and user-friendly experience. • The issues in the model status view have been resolved, providing a more accurate and up-to-date view of the model's status.

1. Add voice activate detection feature in audio file transcription to improve transcription accuracy for long meeting records. 2. Refine the user interface in the live recording view to enhance user experience. 3. Optimize the prompt format and chat template for Large Language Models (LLMs) to achieve more stable and quick results. 4. Implement better model status management to ensure efficient model updates and maintenance. 5. Address UI issues in iOS 26 to ensure compatibility and a smooth user experience.

* TokKong is integrated with CloudKit, which means it can sync all chat history, transcription history across your Apple Devices * Enhance UX * You can also download desktop version from macOS App Store, it allows download custom LLM model

* Improved RAM management for the Whisper engine * Added controls for managing model loading * Enhanced localization support

* Enhance Whisper engine in the recording view * fine-tune the text enhancement feature

- Enhanced audio transcription when using the Whisper engine - Fixed UI issues in the AI chat view

- refactor system prompt, 150% faster than before - polish UI components

* Now supports chatting with WhatsApp exported conversations * Now supports chatting with audio-transcribed text

fix issues

* New Design * Fix issues in AI Chat conversation management

1. Merged memo mode with recoding mode; 2. We upgraded AI Chat Module, so now we have more robust AI conversation; 3. Pro user can download Whisper Models.

1. Added Whisper support — You can now choose between Whisper or iOS native speech recognition for audio transcription. 2. Improved reading mode — Enhanced the reading experience for long texts with better layout and stability. 3. Fixed issues in AI Chat view — Resolved bugs to ensure smoother interactions. 4. Fixed issues in audio transcription view — Improved stability and accuracy in the transcription interface.

Now we support iPadOS!!!

What’s New: * Resolved issue where audio from WhatsApp triggered unwanted playback and caching. * Fixed missing “Clear” button in AI history view. New Features: * Added support for copying and sharing messages in Chat AI mode, now also available in read-only view. * You can now re-process messages with AI directly from the history view for more flexibility.

Enhanced AI Chat System Prompts: Choose from 10+ AI personalities (General Assistant, Day Planner, Math Teacher, etc.) Custom Prompts: Create your own AI personalities Document Mode: Upload documents and ask questions about them Model Management New Models: Added Gemma, Phi-3, Mistral, and Qwen variants Model Dashboard: Easy switching between AI models Storage Monitor: Track and manage device storage AI Controls Response Length: Adjust from 512 to 4096 tokens Creativity: Fine-tune from conservative to very creative Context Size: Configure 2048, 4096, or 8192 tokens Performance Faster model loading and switching Better memory management Enhanced error handling Experience more powerful and personalized AI interactions!

New in this version: Shortcut Support – Make TokKong part of your workflow. We’ve added support for Shortcuts, so you can now integrate TokKong into your daily routine effortlessly. Launch live recording, transcribe voice messages, analyze PDFs, or start memo mode — all with a tap or voice command. Whether you’re on the move or multitasking at work, TokKong now works your way. This update is perfect for business travelers, content creators, and anyone who values speed, privacy, and control — even when offline. Update now and take your productivity to the next level — 100% offline, 100% yours.