Meta is Announced Lalama 4Its latest collection of AI models that now powers Meta AI Assistant On the web on the web And in WhatsApp, Messenger, and Instagram. Two new models, are also available to download From the meta Or The hugs faceLama 4 Scouts-A small model that is capable of “fitting in the same NVIDIA H100 GPU”-and Lama 4 Moric, which is equivalent to GPT-4O and Gemini 2.0 flash. Meta says it is still in the process of training Lama 4 Behimovat, which CEO of Meta Mark Zuckerberg says “The world’s most performing base model.”
According to Meta, Lilama has a 10 million token context window in the 4 scout. Gemini 2.0 Flash Light Models as well as open source wrong behavior 3.1, “in a wide range of widely reported benchmarks,” while still “fits the same NVIDIA H100 GPU.” Meta makes similar claims compared to the performance of her bigger moorous model Openai’s GPT -4 O And Google’s Gemini 2.0 flash, and said that the results are comparable DPSEC-V 3 Coding and reasoning works using “less than half -active parameters”.
Meanwhile, Lama 4 Behimot has 288 billion active parameters with a total of 2 trillion parameters. Although it has not yet been released, Meta says Behmot can overtake its rivals (in this case GPT -4.5 and Claude Swant 3.7) “Several stem benchmarks on the mark.”
For Lalama 4, Meta says it turned into a “mixture of experts” (MOE) architecture, a point that protects resources using only parts of a model that requires a work. The company plans to discuss future plans for AI models and products at its Lamakone Conference, which is Is taking place on April 29.
As its past models, the collection of Meta Lama 4 is called “Open Source”, though Lama has been criticized for licensing restrictions. For example, Call 4 License Commercial companies with more than 700 million monthly active users need to request permission from Meta before using their models, which Open Source Initiative Auto Written in 2023 “Takes it out of the category of open source.”