Meta Platforms has officially launched its much-anticipated Llama 4 AI models, stepping up the competition against OpenAI’s ChatGPT and China’s DeepSeek. Despite earlier delays, the release—announced over the weekend—marks a major leap in Meta’s artificial intelligence ambitions.
According to the company, the newly unveiled models—Llama 4 Scout and Llama 4 Maverick—are designed with powerful multimodal capabilities, meaning they can process not just text, but also images, video, and audio. Developers can now explore these models via Meta’s applications and the Meta.ai website.
Meta touted the models as “best in class,” claiming they outperformed top-tier competitors like OpenAI’s GPT-4o and Google’s Gemini 2.0 Flash in a variety of industry-standard evaluations. Importantly, the models operate as open-weight systems, giving developers more flexibility compared to closed-source alternatives.
The company hinted that its most advanced model is still in the pipeline. A preview of “Llama 4 Behemoth,” described as Meta’s smartest and most powerful language model yet, is in the works and is expected to guide future model training.
However, internal reports suggest that Llama 4’s journey hasn’t been smooth. Reuters, quoting The Information, noted that Meta previously postponed the launch twice due to underwhelming performance in areas like reasoning and mathematics. There are also concerns that Llama 4 still falls short in delivering realistic, voice-based interactions compared to OpenAI’s offerings.
Despite these hurdles, Meta remains committed to AI dominance. CEO Mark Zuckerberg, during the company’s Q4 2024 earnings call, said Llama 4 comes with “agentic” intelligence and native multimodal features that will open the door to a wave of new applications.
To support its AI expansion, Meta is planning to invest between $60 billion and $65 billion in 2025 to scale up its AI infrastructure.
