Official announcements from AI companies and reputable AI news sources.

Meta Releases Llama 3.1 405B AI Model on July 23, 2024, Outperforms GPT-4o

Jul 23, 2024, 03:08 PM

Meta has officially released its highly anticipated Llama 3.1 AI model on July 23, 2024, including the 405 billion parameter version, which is the largest open-source model to date. The Llama 3.1 405B model has been leaked on platforms like 4chan and Reddit prior to its official release, showcasing benchmarks that outperform OpenAI's GPT-4o and Anthropic's Claude Sonnet 3.5 in several metrics. The model is trained on 15 trillion tokens and supports multilingual capabilities, with a context length of 128K tokens. Meta's Llama 3.1 also includes improved versions of their 8B and 70B models. The release marks a significant milestone in the AI community, as it is the first open-source model to rival top proprietary models. Mark Zuckerberg highlighted the significance of this release in an interview.