Llama is Meta's advanced open-source AI model family. The latest Llama 4 features a mixture-of-experts architecture, supports 10 million token context length, native multimodality, and matches GPT-4 performance. Freely available for download, it empowers developers worldwide to build innovative AI applications without vendor lock-in.