LocalMind: Ollama & LM Studio

NEW FEATURES & ENHANCEMENTS: • Apple MLX Integration: Experience lightning-fast, fully local AI inference on Apple Silicon hardware with Metal acceleration, streaming, and full model lifecycle management. • Home Screen Widgets: Launch new chats, access voice mode, or start a search immediately from your home screen with seamless deep linking. • Built-in System AI Support: Added compatibility with system AI models alongside hardware capability checks and localized guidance. • Reasoning Mode & Thinking Controls: Fine-tune model thinking efforts across local GGUF models, Ollama, and LM Studio with per-model options (Default / On / Off). • Location-Aware AI Tools: Enable AI models to use device location and geocoding services with your permission for weather, directions, and local queries. • New Languages: Added complete Portuguese and Korean localizations with an improved language selector during onboarding and in settings. • System TTS Customization: Better audio synthesis supporting system speech engines with automatic discovery, without overriding custom preferences. PERFORMANCE & STABILITY FIXES: • MCP Tool Completion Loop: Resolved an issue where tool execution results were not sent back to the model, ensuring multi-step tools can be completed seamlessly. • Multi-Turn Conversation Memory: Fixed multi-turn context retention for local on-device models. • Voice Mode & Mic Stability: Enhanced microphone initialization, permission handling, and background audio continuity. • Memory Optimizations: Implemented session idle TTL cleanup to prevent background memory overhead during inactive conversations. • Crash Preventions: Addressed screen lifecycle and drawer navigation edge cases across the app for a smoother, more reliable experience.