Cactus: Cross-Platform Framework for Local LLM Deployment

Cactus is a cross-platform framework for deploying large language models (LLMs), vision language models (VLMs), and text-to-speech (TTS) models locally within your app. Supporting Flutter and React Native, it works with any GGUF model from Hugging Face (Qwen, Gemma, Llama, etc.), handling models from FP32 down to 2-bit quantization. Cactus provides MCP tool calls for enhanced AI functionality (reminders, image search, message replies), cloud model fallback for complex tasks, Jinja2-powered chat templates, and token streaming. Example code, performance benchmarks across various devices, and C++ backend are provided.