Meta has released its first models from the latest open-source AI suite, Llama 4, with a strategic push to strengthen its foothold in the generative AI space. The initial two models, Scout and Maverick, are available for download via the Llama website and Hugging Face.
The Llama 4 models are designed to operate on a single GPU, reflecting a trend towards lightweight high-performance models. Meta claims that Maverick outperforms OpenAI’s GPT-4o and Google’s Gemini 2.0 Flash across several benchmarks, including code generation, reasoning, and multilingual understanding.
Llama 4 Behemoth, still in development, is expected to feature 288 billion active parameters and nearly 2 trillion total parameters. The model is designed to help train and guide future models.
Meta’s CEO Mark Zuckerberg has stated that the company aims to build the world’s leading AI, open-source it, and make it universally accessible. The Llama 4 rollout also marks Meta’s first use of a mixture-of-experts framework, which boosts efficiency and reduces costs for training and inference.
The release comes after downloads of previous Llama models surpassed one billion, with Meta investing $60-65 billion in AI infrastructure spending for 2025. With the launch of Llama 4, Meta is solidifying its position as a major player in the generative AI space.
Source: https://www.ndtv.com/world-news/meta-launches-llama-4-all-about-the-latest-open-source-ai-model-8100928