Meta to launch Llama 4 AI model this year

Meta is set to release its next-generation AI model, Llama 4, later this year. This model is anticipated to significantly enhance the capabilities of AI technology. Llama 4 is expected to feature new reasoning abilities and allow AI agents to access the internet and use various tools. The company previously released Llama 3.3 in December 2023, which greatly improved performance and reduced costs. CEO Mark Zuckerberg stated that training Llama 4 will require ten times more computing power than was needed for Llama 3. He emphasized the importance of building this capacity ahead of demand. Llama 4 will include "agentic capabilities," meaning it can perform tasks like a human engineer, rather than just responding to commands. This development could lead to AI being able to handle complex jobs independently. Clara Shih, head of business AI at Meta, noted that many businesses will soon use AI to automate repetitive tasks and enhance customer interactions. However, Zuckerberg warned that the widespread use of autonomous AI tools won't happen until around 2026. He believes that this year will lay the groundwork for future advancements. As the technology evolves, it could also drive down costs and improve access for developers. To support these advancements, Meta plans to invest up to $65 billion this year in AI infrastructure. The company is building a new 2-gigawatt AI data center to accommodate the increased demands of future AI models. The goal is to create proactive AI that can perform tasks autonomously, which will ramp up competition in the AI industry.