Categories
Meta Launches Llama 4 AI Model Series
Meta Platforms has launched its most up-to-date AI model, Llama 4 Scout and Llama 4 Maverick which mark a major advancement of multimodal AI. These models process and incorporate different types of data comprising text, images and videos, further enhancing their abilities across multiple applications.
Llama 4 Scout is developed to be efficient and can run with just the one Nvidia H100 GPU. It’s able to provide the largest context window, which can handle greater than ten million tokens. This makes it able to process large amounts of data quickly. This system is especially good at long-term contextual tasks, like analysing multiple documents or large code bases.
Meta has opened both of its models to the public, showing its dedication to collaboration on AI development. Meta is integrating them in platforms such as WhatsApp, Messenger, and Instagram to increase user interactions.
Alongside Scout and Maverick, Meta is currently working on Llama 4 Behemoth, which is a model that has more than 288 billion active parameters and two trillion in total. Behemoth aims to surpass current STEM benchmarks and contribute to the training of the upcoming AI models.
Meta’s launch in the Llama 4 series underscores its commitment to developing AI technology while also providing easy-to-use, high-performance models to the wider community.
For more information regarding Meta’s AI developments, be sure to check atoztechworld.
