Meta's open source large model Llama 4 is released, Alibaba/WiMi focuses on AI applications in vertical fields

Llama 4: Meta's first natively multimodal mixture-of-experts (MoE) architecture models

A little over a week ago, Meta announced the release of its new-generation Llama 4 models, marking what the company calls "the beginning of a new era of natively multimodal AI innovation." The new suite includes Scout and Maverick, two immediately available models and previews Behemoth, a larger, more powerful model still in development.Technical InnovationsThe newly released models feature Meta's first implementation of mixture-of-experts (MoE)…

Meta's open source large model Llama 4 is released, Alibaba/WiMi focuses on AI applications in vertical fields

On April 6, Beijing time, the US technology giant Meta (META) launched the open source artificial intelligence model Llama 4. According to reports, the model currently has two versions, Scout and Maverick. It is Meta’s most advanced model to date and the most multimodal model among similar products. The latest AI large model Llama 4 is unveiled Meta said in a statement that Llama 4 is a multimodal AI system that can process and integrate multipl…

Hi everyone, I'm Vincenzo Lomonaco, researcher and professor at the University of Pisa. In today's episode, we'll talk about the recent release of Meta AI's highly anticipated open source model, Llama 4, and the incredible technological innovations it introduces. Is this really a leap forward in terms of quality compared to the models currently in vogue? What [...] The article The Voices of AI — Episode 96: LLama 4: Meta.ai's new ecosystem of op…