Meta Platforms Releases New AI Models

On Saturday, Meta Platforms unveiled two new advanced models in its large language model (LLM) lineup: Llama 4 Scout and Llama 4 Maverick. These models aim to enhance the capabilities of AI through multimodal systems.

As a multimodal AI system, Llama is designed to process and integrate various data types, including text, video, images, and audio, allowing it to convert content across these formats. Meta stated that Llama 4 Scout and Llama 4 Maverick represent “the best in their class for multimodality.”

Open Source Software Initiatives

Both Llama 4 Maverick and Llama 4 Scout will be released as open-source software. Additionally, Meta is previewing Llama 4 Behemoth, described as “one of the smartest LLMs in the world” and serving as a teaching tool for subsequent models.

AI Development Landscape

In recent years, major tech companies have significantly invested in artificial intelligence infrastructure, particularly following the success of OpenAI’s ChatGPT, which transformed the tech landscape and prompted an increase in machine learning investments.

However, reports indicate that the launch of Llama 4 was delayed due to concerns that its initial development fell short of Meta’s technical benchmarks in reasoning and math.

Continued Investment and Future Directions

According to sources, Meta plans to invest as much as $65 billion this year to further bolster its AI infrastructure, responding to investor pressure for substantial returns on these investments.

• 0 Comments
• Llama 4