Meta is ready to offer their artificial intelligence model, Llama in open source form – thus enabling many to develop their own artificial intelligence solutions, using Meta's vernacular model. Most recently, in parallel with various partnerships, Mark Zuckerberg also shared that Meta is currently training the Llama 4 model, with the target of introduction in early 2025.
This Llama 4 model is being trained using larger clusters than the 100,000 H100s, or what has been reported so far. Previously, Elon Musk has said they have the largest AI training cluster in the world, with 100,000 NVIDIA H100s. Higher computing power is seen as needed to train AI models, making them smarter and more capable.
For now, it has not been shared about the new features and additions that will be introduced through Llama 4. Mark Zuckerberg simply stated that it will have better reasoning support, in addition to operating faster.