Meta has unveiled Llama 3.1, its most advanced generative AI model to date. Known as Llama 3.1 405B, this model is touted as the world’s largest and most capable foundation model available to the public, competing with leading models from OpenAI and Google.
AI Model Capabilities
Meta claims Llama 3.1 405B excels in general knowledge, steerability, mathematics, tool use, and multilingual translation. The model supports a larger 128,000-token context window and can respond in multiple languages, including French, German, Hindi, Italian, Portuguese, and Spanish.
Integration Across Meta Products
Since the release of Llama 3 in April, Meta has integrated generative AI into its products, such as Instagram, Messenger, and WhatsApp.
CEO Mark Zuckerberg emphasises the importance of open-source AI, advocating for widespread access to AI technology.
Open Source and Industry Impact
Meta’s commitment to open-source AI has influenced other developers, including Venice AI, Brave, and Perplexity AI, to adopt Llama 3.
Venice AI founder Erik Voorhees praised Meta for its contribution to the AI community by providing a cutting-edge model for free.
Zuckerberg talks about artificial intelligence
Zuckerberg addresses AI safety, distinguishing between unintentional and intentional harm. He highlights the transparency of open-source software as a mitigating factor for AI risks. Meta’s new tool, Llama Guard 3, aims to monitor and enforce ethical use of the model.
Zuckerberg believes the release of Llama 3.1 marks a pivotal moment in the AI industry. He asserts that open-source AI offers the best chance to maximise economic opportunities and security for all.