Hugging Face has integrated MLX LM, enabling users to run over 5,000 large language models (LLMs) locally on Apple Silicon devices at maximum speed without relying on cloud services. This integration allows Mac users to access these models directly through the Hugging Face Hub by simply clicking "Use this model." Additionally, users can now spin up OpenAI-compatible servers directly from the model page. Separately, WebLLM has been introduced as a browser-based solution that supports full GPU acceleration and OpenAI API compatibility, allowing LLMs to run directly in the browser with features such as streaming, JSON-mode, and function-calling.