Meta’s Llama 4-based next-generation artificial intelligence model “Behemoth” will miss its launch date once again. The model, which was initially planned to be introduced at the LlamaCon event held in April, was later postponed to June.

Meta has officially postponed its “Behemoth” artificial intelligence

According to recent developments, Behemoth’s launch has been postponed to the fall or a later date. It is said that Meta engineers are having difficulty increasing the model’s performance to the expected level. Behemoth, which Mark Zuckerberg describes as “the world’s highest-performing model,” will have a total of 2 trillion parameters, 288 billion of which are active.

In its evaluations last month, Meta announced that this model had outperformed its competitors such as GPT-4.5, Claude 3.7 Sonnet, and Gemini 2.0 Pro in many STEM tests. However, despite these statements, it is still unclear when the final version of the model will be available.

Meta has so far introduced two small-scale models of the Llama 4 family, Scout and Maverick, to users. It is also stated that a lighter model codenamed “Little Llama” is on the way.

However, the delay of Behemoth, which the company has been working on for a long time, has begun to question Meta’s claims of leadership in the field of artificial intelligence. While only superficial information about Behemoth was shared at the last event, no concrete statement was made about the current status of the model or its exact release date.

Meta plans to invest $72 billion in artificial intelligence infrastructure in 2025. So what do you think about this? You can share your views with us in the comments section below.