V6: Local AI, Image Generation, and OCR
Local AI. Image generation. OCR. Diagram editing. And a VS Code-style interface. V6 is here.
Six weeks ago, we shipped V1. Now we're at V6—and this release takes Diwadi to a whole new level. Run AI completely offline with local LLMs, generate images with Stable Diffusion, extract text from images, and edit diagrams—all without leaving the app.
What's New in V6
Local LLM Support
Run AI models completely offline using llama.cpp. Full tool calling support so the AI can interact with your files. Auto cleanup when idle to save resources. Your data never leaves your machine—true privacy-first AI.
AI Image Generation
Generate images from text prompts using Stable Diffusion. Choose from available models, customize settings, and create images instantly. Automatic process management handles everything behind the scenes.
OCR & Diagram Editing
Extract text from any image with built-in OCR. Edit D2 diagrams with live rendering. Preview Mermaid diagrams in real-time. Export Typst documents to PDF. Support for Intel Macs included.
VS Code-Style Title Bar
A completely redesigned title bar inspired by VS Code. Revamped split views for better multitasking. New notification system that's less intrusive. The app feels more professional than ever.
Even More Features
7-Zip compression support. Apple Sign-In for seamless authentication. Subtitle search and replace. Improved PPTX and DOCX handling. Unified drag and drop across the entire app.
Windows-Ready Builds
V6 brings improved Windows support with ready-to-ship builds. Cross-platform compatibility is better than ever—same great experience on macOS, Windows, and Linux.
The Technical Details
Complete llama.cpp integration with tool calling. Stable Diffusion pipeline with automatic model management. OCR powered by efficient local processing. D2 and Mermaid diagram rendering. Typst to PDF export. Unified drag-drop system across all views.
The Numbers
120+ commits since V5. 20+ bug fixes. Three AI engines (cloud, local LLM, image generation). All while keeping the app lightweight and fast.
Six Weeks of Shipping
V1 on Oct 31. V2 on Nov 7. V3 on Nov 14. V4 on Nov 21. V5 on Nov 28. V6 today. We're building in public and shipping every week. This is how you build software that people actually want.
View Full Changelog →Six releases. Six weeks. Local AI. Image generation. Download V6 now and experience the future of desktop productivity.
Download Diwadi V6Full Changelog
Major Features
- Local LLM support with llama.cpp—run AI models completely offline
- Full tool calling support for local LLMs
- Auto cleanup when LLM is idle to save resources
- AI image generation with Stable Diffusion integration
- Multiple model support for image generation
- OCR text extraction from images
- D2 diagram rendering and editing
- Mermaid diagram live preview
- Typst preview and PDF export
- 7-Zip compression support
- Apple Sign-In authentication
Technical Improvements
- llama.cpp integration with proper process management
- Stable Diffusion pipeline with automatic model downloads
- OCR with auto compression for better performance
- Intel Mac support for OCR features
- Windows-ready builds with improved compatibility
- Unified drag-drop system across all views
- Improved PPTX and DOCX handling
UI/UX Enhancements
- VS Code-style title bar design
- Revamped split views for better multitasking
- New notification system
- Subtitle search and replace functionality
- Improved drag-and-drop feedback
- Better loading states throughout the app