Meta releases new versions of its Llama large language model, called Llama 4 Scout and Maverick.
The Llama 4 large language model (LLM) is described as a multimodal AI system, meaning it can process and integrate different types of data such as text, video, images, audio, and convert content between these formats.
Meta claims that Llama 4 Scout and Llama 4 Maverick are its “most advanced models,” “leading the multimodal intelligence segment,” and are open-source software.
Facebook's parent company also released a preview of Llama 4 Behemoth, which it bills as "one of the smartest LLMs in the world" and will serve as a "teacher" for future models.
Major tech companies are investing heavily in AI infrastructure following the success of ChatGPT, a game-changer in the tech industry and a boost in machine learning investments.
Meta delayed the launch of Llama 4 due to its failure to meet the company's technical standards, especially in tasks involving reasoning and mathematics. The company was concerned that Llama 4 would not be as effective as OpenAI's models at conducting human-like voice conversations.
Meta plans to spend $65 billion by 2025 to expand its AI infrastructure, as investors pressure big tech companies to demonstrate the effectiveness of their huge spending.