v1.2.0

## [1.2.0] - 2025-09-10

### ✨ Features

**Native SafeTensors Support**
- Implemented native SafeTensors inference engine with zero Python dependencies
- Added complete SafeTensors model format support alongside GGUF
- Enhanced model detection and loading for SafeTensors files

**Enhanced Build System**
- Updated release workflow with comprehensive system dependencies
- Added support for all-features builds across platforms
- Improved cross-platform compilation reliability

### 🐛 Bug Fixes

**Build and Deployment Issues**
- Fixed release binary generation to exclude problematic llama.cpp dependencies
- Resolved macOS runner cmake installation conflicts
- Enhanced GitHub Actions workflow with proper dependency management

**Model Discovery Improvements**
- Fixed infinite recursion issues in model discovery on macOS
- Enhanced model loading robustness across different file formats
- Improved error handling for corrupted or incomplete model files

### 🚀 Enhancements

**Performance Optimizations**
- Native SafeTensors processing for faster model loading
- Reduced memory footprint with optimized inference pipeline
- Enhanced startup performance with efficient model detection

**Developer Experience**
- Comprehensive testing suite for SafeTensors functionality
- Improved documentation for multi-format model support
- Enhanced debugging and error reporting capabilities