Meta introduces Llama 4 with two new AI models available now, and two more on the way

Meta, Multimodal Llama 4 left the first two models of Suite: Llama 4 Scout and Llama 4 Maverick. Maverick, the concept of imagery and text for “common assistance and chat use cases” is the concept of text for “general assistants and conversations”, “he said. Blog PostSmall Model Scout “Many documentaries like multi-documentary generalization, extensive user activities and justification over large codes for individual tasks”. “It can solve things like. The company also said that a client says that 4 Behemoth,” The world’s smartest LLS “is also CEO Mark Zuckerberg The fourth model, Llama 4, said we will hear about “in the next month.”

Both Maverick and Scout are now available to download Llama Website and Hug faceAnd WhatsApp, Messengram and Instagram added to Meta AI including DMS.

Scout has 16 specialists, there are 17 billion branch parameters, says Meta. According to Zuckerberg, “It is a very fast, justified multimodal and an industry owner, about 10 million Token context length and designed to work in a GPU.” On the other hand, Maverick is a 17 billion active parameter with 128 specialists. The company defeated opponents such as GPT-4O and Twins 2.0 on Multilingual, Reasoning, Multilingual, Long-Term and View Criteria and accumulated opponents against DeepSEEK v3.1 during mind and coding.

According to the company, Zuckerberg calls on 288 billion active parameters, “The world’s highest performance base model”, which is still the approaching Behemoth model. It may not be here yet, but probably we will be able to hear more and confident model soon; Meta’s Great EU Developer Conference, LlamaconIt’s been a few weeks.