AI Marches On [English]

Author: Shelly Palmer | Source: Shelly Palmer | Read the full article in English

AI technology continues to evolve rapidly, with Meta Platforms recently announcing the launch of Llama 4, their latest series of large language models. This new release includes two models: Llama 4 Scout, which is designed to be efficient and run on a single Nvidia H100 GPU, and Llama 4 Maverick, which offers strong performance in coding and reasoning tasks. Both models are capable of handling various types of data, such as text, images, and audio, making them versatile for different applications.

In addition to these models, Meta is working on an even more advanced version called Llama 4 Behemoth, which boasts an impressive 288 billion active parameters. This model is expected to outperform other leading models in STEM-related tasks. All Llama 4 models utilize a unique architecture that optimizes their performance by activating only the necessary components for specific tasks, which helps in managing resources effectively.

Meta's ambitious plans include integrating these AI models into their various platforms, such as WhatsApp and Instagram, with a goal of reaching one billion users by the end of 2025. To support these initiatives, the company has committed a substantial investment of up to $65 billion for the year, highlighting their dedication to advancing artificial intelligence technologies.

[Read More (translated)]