Meta Unveils Llama 4 Maverick and Scout, Boosts AI Ambitions

Meta Platforms (NASDAQ:META) has unveiled its latest AI models—Llama 4 Maverick and Llama 4 Scout—calling them its “most advanced” large language models (LLMs) to date. Designed as multimodal AI systems, they can process and integrate diverse data formats including text, images, video, and audio, pushing Meta’s capabilities in generative AI further than ever.

Both models are being released as open-source software, reinforcing Meta’s commitment to collaborative AI development. The tech giant also previewed Llama 4 Behemoth, a forthcoming model described as “one of the smartest LLMs in the world,” intended to serve as a training foundation for future models.

Meta emphasized that these models set new standards for multimodal performance, aiming to rival or surpass competitors in the fast-moving AI space. However, a recent report from The Information suggested Meta previously delayed the launch of Llama 4 due to underperformance in technical benchmarks, especially in math and reasoning tasks. Concerns were also raised about Llama 4’s limitations in conducting humanlike voice conversations, an area where OpenAI’s ChatGPT continues to dominate.

In response, Meta is aggressively scaling its infrastructure, planning to invest up to $65 billion this year in AI hardware and systems. With mounting investor pressure, the company is striving to prove its ability to convert AI innovation into tangible returns.

As the race among big tech companies to lead in artificial intelligence intensifies, Meta’s Llama 4 lineup signals a strong push to reclaim ground and shape the future of AI through open access and cutting-edge multimodal capabilities.