The first models from Meta’s most recent open-source AI program, Llama 4, were made public on Saturday as the business tries to take the lead in generative AI investments.
However, the owner of Facebook claimed that the Llama 4 model, which is the largest and most potent of its kind and acts as “a teacher for our new models,” has not yet been made public. The business stated in its blog post that the so-called Llama 4 Behemoth model is currently undergoing training.
Chris Cox, the chief product officer of Meta, stated in March that Llama 4 will assist in powering AI agents, who will be able to think and act at new levels. These agents will have the ability to browse the internet and do a variety of jobs that may be useful to both businesses and consumers.
Two recently published Llama 4 models on Meta AI are available for users to test out on the Meta AI website, Instagram Direct, WhatsApp, and Messenger. Llama 4 Scout and Llama 4 Maverick are included in that.
In an Instagtam video, Meta CEO Mark Zuckerberg stated, “Our objective is to create the most advanced AI in the world, make it open source, and make it available to everyone so that everyone in the world benefits.”
“And with Llama 4, I’m beginning to see that open source AI will become the dominant models, as I’ve been saying for a while,” he continued. “A major update is being made to Meta AI today.”
On April 29, Meta will hold its inaugural LlamaCon AI conference. According to a February CNBC article, the company is also anticipated to reveal a stand-alone app for its Meta AI chatbot in the second quarter.
