Meta Platforms has released the latest version of its large language model, introducing Llama 4 Scout and Llama 4 Maverick. According to Meta, these are the company’s most advanced models yet, designed for multimodal AI—capable of processing and integrating text, images, audio, and video.
Both models will be released as open source software, further supporting Meta’s strategy to promote transparency and innovation in AI development. Additionally, the company previewed Llama 4 Behemoth, which it described as “one of the smartest LLMs in the world” and its most powerful model to date, intended to help train future AI systems.
This launch comes amid a broader push by major tech firms to ramp up AI capabilities following the success of OpenAI’s ChatGPT. In response, Meta plans to invest up to $65 billion this year to grow its AI infrastructure.
However, development of Llama 4 reportedly faced delays. According to The Information, the models initially failed to meet Meta’s technical benchmarks—particularly in reasoning, math, and humanlike voice conversations—leading to a postponement in the release schedule.
