Meta Platforms (META) may release a new large language model in the first-quarter of 2026, as the Mark Zuckerberg-led company looks to further compete with Google (GOOG) (GOOGL), OpenAI (OPENAI) and others, CNBC reported, citing people with knowledge of the matter.
The new model, code-named Avocado, is being seen as a successor to Llama, which Meta has had struggles with, the news outlet added. When Avocado is released, it is likely to be a proprietary model, the people added. Meta’s family of Llama models are currently open-sourced, which lets anyone download them and improve them.
In June, it was reported that Meta executives, including Zuckerberg, had discussed “de-investing” in its Llama family of language models in favor of models made by competitors, including those made by OpenAI and Anthropic (ANTHRO).
A couple of months later, Meta restructured its artificial intelligence division, in part to better organize the company so it can develop A.I. products more quickly to compete with others.
This summer, Meta paid nearly $15B to acquire a stake in Scale AI and make its CEO, Alexandr Wong, Meta’s Chief AI Officer.
Meta did not immediately respond to a request for comment from Seeking Alpha.