The user wants a macOS version of the Lekh AI app, bringing the same on-device AI experience to desktop.
Hey everyone I’m the solo developer behind **Lekh AI**, an on-device AI app for iPhone & iPad. I just shipped **v2.0**, and this release is focused on making local models more flexible, faster, and more reliable. **Quick recap:** Lekh AI runs LLMs, vision, image generation, and voice **entirely on-device**. No cloud. No accounts. No subscriptions. Your data stays on your device. **What’s new in v2.0** **LLaMA GGUF support** * Load and run **GGUF LLaMA models** locally * Much better compatibility with community models * Easier experimentation with different model sizes **Better RAG memory** * Improved recall and relevance * More consistent use of stored context across chats * Fewer “why did it forget that?” moments **TTS optimizations** * Faster, smoother voice output * Reduced latency and improved stability in longer sessions **UX & cleanup** * Removed the persistent uncensored-model warning * Cleaner model switching experience * General polish across the app **Bug fixes & performance improvements** * Fewer hiccups during long chats * Better memory management * Overall smoother feel **Smarter AI & Memory** * Custom AI personas (role-consistent, persistent) * View, edit, and fine-tune RAG memories * Chat summarization * Better RAG integration across chats * Ask the AI about your book progress directly in chat **New AI Image Tools (all offline)** * AI image editing with **SD 1.5 inpainting** * Ability to load custom models as well * Object remover * Black & white photo colorizer * Photo → 3D depth generation * 3D splat generator + viewer * Image editing now feels way more “Photos-app-like” **Documents & Reading** * Improved document & PDF handling * Better long-file performance * More reliable book context awareness **Plus:** lots of bug fixes and stability improvements **Core features (for anyone new)** * Offline LLM chat (Gemma, Qwen, Llama, Mistral, Phi, DeepSeek, OpenELM, more) * Vision: ask questions about images and photos * On-device image generation (SD 1.5 / SDXL) * Voice chat with Kokoro TTS * Local AI server (OpenAI-compatible API over LAN) * iCloud sync (optional, encrypted) * **One-time price: $4.99 - no subscriptions** **What’s next**: * **macOS app ships next week**, bringing the same fully on-device experience to desktop **App Store link:** [https://apps.apple.com/us/app/lekh-ai/id6757496953](https://apps.apple.com/us/app/lekh-ai/id6757496953) I’m building this very openly, and feedback genuinely shapes the roadmap. If you’re into **local AI, privacy-first apps, or running models on Apple devices**, I’d love to hear what you think 🙏 Happy to answer any technical questions in the comments. [](https://www.reddit.com/submit/?source_id=t3_1qys3r7)