Meta has unveiled Llama 3.3 70B, the newest addition to its Llama generative AI model family. According to Ahmad Al-Dahle, VP of generative AI at Meta, this text-only model matches the performance of Meta’s larger Llama 3.1 405B model but at a lower cost, thanks to advanced post-training techniques. The model outperforms competitors like Google’s Gemini 1.5 Pro, OpenAI’s GPT-4o, and Amazon’s Nova Pro on several benchmarks.
Llama 3.3 70B is available for download from Hugging Face and the official Llama website. While Meta imposes usage restrictions on platforms with over 700 million monthly users, the model has still seen widespread adoption, with over 650 million downloads.
Internally, Meta’s AI assistant, powered by Llama models, now boasts nearly 600 million monthly active users. Meta is also addressing technical challenges by building a $10 billion AI data center in Louisiana and expanding its computing infrastructure with over 100,000 Nvidia GPUs. However, Meta faces regulatory hurdles, particularly with the EU’s AI Act and GDPR compliance, affecting its data training practices.
