Meta debut new Lama four fashions, however probably the most highly effective AI mannequin will nonetheless come

Visitors take photos with a sign that was published on January 29, 2025 in front of the Meta headquarters in Menlo Park, California.

Justin Sullivan | Getty pictures

Meta The first models from its latest open source artificial intelligence software Llama 4 released the first models on Saturday when the company headed the race for investing in generative AI.

But the Facebook owner said that he has not yet published the largest and most powerful Lama 4 model, which other AI models surpassed in his class and serves as a “teacher for our new models”. According to the company's blog post, this so-called LAMA 4 giant model is still in training.

Llama 4 will help with the Power -Aii agent who will be able to make a new degree of argument and measures, said Chris Cox, Chief Product Officer from Meta, in March. These agents will be able to surf the web and do many tasks that could be useful for consumers and companies.

Users can try out two newly published Llama 4 models on Meta AI in WhatsApp, Messenger, Instagram Direct or the Meta AI website. These include Lama 4 Scout and Lama 4 Maverick.

“Our goal is to build up the world's leading AI, make open source IT and make them universally accessible so that everyone in the world benefits,” said Meta -CEO Mark Zuckerberg in a video about Instagtam.

“And I said for a while that I think Open Source Ki will be the leading models, and Lama 4 will happen,” he said. “Meta Ai gets a big upgrade today.”

Meta will organize its first Lamacon AI conference on April 29. The company is expected to announce an independent app for its META AI chat bot in the second quarter, reported CNBC in February.