Meta has rolled out its latest Llama 4 AI models, introducing three distinct versions — Scout, Maverick, and Behemoth — as part of a broader global strategy to strengthen its role in the open-source AI arena. The models, launched on Saturday, are built on a mixture-of-experts (MoE) architecture, designed to enhance performance by routing tasks through specialized sub-networks.
Maverick, the centerpiece of the lineup, is positioned as a general-purpose AI assistant and has reportedly outperformed OpenAI’s GPT-4o and Google’s Gemini 2.0 on key benchmarks. However, it still trails behind top-tier systems like GPT-4.5 and Gemini 2.5 Pro, according to initial evaluations by TechCrunch.
Also Read: Book Criticizing Meta Tops Bestseller List
The models are freely accessible through Meta’s website and open-source platforms such as Hugging Face — though with a significant caveat. Developers and companies within the European Union are blocked from accessing them, a decision Meta attributes to the EU’s stringent AI regulations under the Artificial Intelligence Act.
Meta has been vocal in criticizing the EU’s regulatory approach, describing the compliance demands as stifling to innovation. The exclusion of EU-based users reflects the growing friction between advancing AI capabilities and maintaining legal and ethical safeguards.
With this latest release, Meta underscores its ambition to remain a dominant force in the AI space while highlighting the mounting challenges tech giants face in balancing innovation with global regulatory pressures.
