Meta Unveils Llama 4 AI Models

Meta Platforms has released the latest iteration of its large language model, introducing Llama 4 Scout and Llama 4 Maverick as part of its expanding AI ecosystem. The company describes the models as its most advanced yet, showcasing capabilities in multimodality—allowing them to process and integrate diverse data formats including text, audio, video, and images. Both models will be available as open-source software, reflecting Meta’s continued commitment to openness in AI development.

Alongside the release, Meta also offered a preview of Llama 4 Behemoth, which it claims to be one of the smartest large language models in the world. Behemoth is expected to serve as a foundational system to further enhance the intelligence of upcoming models. While the new releases represent a major milestone, internal reports suggest that Meta delayed the launch due to underperformance in certain technical benchmarks, particularly in reasoning and mathematical tasks. Additionally, early tests indicated that Meta Llama 4 was not as advanced in humanlike voice interaction as OpenAI’s flagship models.

Despite these challenges, Meta is doubling down on its AI ambitions. The company is planning to invest up to $65 billion this year to strengthen its AI infrastructure—a move aimed at keeping pace with competitors like OpenAI and Google DeepMind. This investment includes expanding data centers, acquiring more computing resources, and supporting research efforts to improve model performance. With Llama 4 now entering the public domain, Meta is positioning itself to be a leading force in the rapidly evolving landscape of generative AI.