Meta has introduced the fourth generation of its Llama artificial intelligence models, launching three distinct versions—Scout, Maverick, and Behemoth—as part of its expanded effort to lead in the open-source AI space.
Announced on Saturday, the new models are built on a mixture-of-experts (MoE) architecture, which enhances performance by routing tasks through specialized segments of the model, improving both speed and efficiency.
Meta says the Llama 4 suite is optimized for high-level tasks including reasoning, coding, and multimodal functions such as processing images and videos.
Among the trio, Maverick stands out as a versatile, general-purpose AI assistant. Meta claims it outperforms OpenAI’s GPT-4o and Google’s Gemini 2.0 in several industry-standard benchmarks. However, according to TechCrunch, Maverick still trails behind newer, more powerful models like GPT-4.5 and Gemini 2.5 Pro.
The Scout and Maverick models are now available to the public via Meta’s official platforms and AI distribution hubs such as Hugging Face. However, access is currently restricted for developers and organizations within the European Union. Meta cited the EU’s stringent AI regulations—specifically the Artificial Intelligence Act—as a barrier, arguing that its requirements stifle innovation.
The company has been vocal in its opposition to the legislation, describing compliance as overly burdensome and potentially damaging to global AI development.
With this rollout, Meta is doubling down on its ambition to remain a global frontrunner in AI, even as regulatory environments become increasingly complex. The launch underscores the growing tension between rapid AI innovation and the evolving legal frameworks designed to ensure privacy, transparency, and accountability.
Meta’s aggressive expansion also signals heightened competition among tech giants to define the next era of generative AI, especially within the increasingly influential open-source community.
