Social media giant Meta Platforms (META, Financial) is facing a crisis of market confidence due to delays in the development of its flagship AI model, "Behemoth." The model, part of the Llama 4 series, was initially expected to launch in June but has been postponed to later this year due to technical hurdles encountered by engineers. This model, featuring a cutting-edge Mixture of Experts (MoE) architecture, is crucial to Meta's next-generation AI strategy.
Derived versions of the model, Maverick and Scout, are already available to developers, showcasing the model's ability to handle text, images, and videos simultaneously, outperforming several leading competitors in benchmark tests. However, Meta's AI advancement is under dual pressure: intensified competition with Chinese tech companies like Alibaba (BABA), Tencent, and Baidu (BIDU), and the need to balance technological breakthroughs with cost management.
The delay highlights the broader challenges faced by large tech firms in the AI arms race. While Meta launched the Llama 4 series in April, commercializing the core model remains a challenge due to engineering issues. Insiders report that the team is focusing on improving the model's stability and output consistency, which did not meet expectations in early tests.
