In April, Meta Platforms revealed that it was working on an open-source AI model that performed as well as proprietary models from top AI companies such as OpenAI and Anthropic. Now, Meta CEO Mark Zuckerberg says that model has arrived in the form of Llama 3.1 405B, “the first frontier-level open-source AI model.” The company is also releasing “new and improved” Llama 3.1 70B and 8B models. In addition to general cost and performance benefits, the fact that the Llama 3.1 405B model is open source “will make it the best choice for fine-tuning and distilling smaller models,” according to Meta.

“Llama 3.1 is significantly more complex than the smaller Llama 3 models that came out a few months ago,” writes The Verge, explaining that the largest, Llama 3.1 405B, “has 405 billion parameters and was trained with over 16,000 of Nvidia’s ultra expensive H100 GPUs.”

While Meta hasn’t shared the cost of developing Llama 3.1, “based on the cost of the Nvidia chips alone, it’s safe to guess it was hundreds of millions of dollars,” The Verge reports.

In a Meta newsroom post, Zuckerberg writes that the latest Llama 3 models are “the next steps towards open-source AI becoming the industry standard,” adding the claim that “Llama is already leading on openness, modifiability, and cost efficiency.” By 2025 Meta expects Llama will be the industry’s most advanced AI in terms of performance.

ZDNet details “three decisions” the Meta AI scientists made in building Llama 3.1 that break with conventional approaches and constitute “a tour de force of neural network engineering” and efficiency.

The open source nature of Llama 3 means Meta is making access available for free, including for commercial applications, under the company’s Acceptable Use Policy.

This latest Llama 3 announcement “underscores the social network’s massive investment in keeping up in AI spending with the likes of highflying startups OpenAI and Anthropic and other tech giants like Google and Amazon,” CNBC reports.

“The announcement also highlights the growing partnership between Meta and Nvidia,” notes CNBC, describing Nvidia as “a key Meta partner” that is providing the pioneering platforms with ample supplies of the GPUs necessary to build and train AI models at a time when companies are in competition to buy them.

In what is being billed as a rare joint public appearance, Zuckerberg and Nvidia CEO Jensen Huang will discuss artificial intelligence and virtual worlds during a SIGGRAPH keynote session on July 29.

Related:
Meta’s New Llama 3.1 AI Model Is Free, Powerful, and Risky, Wired, 7/23/24
Meta Llama 3.1 Is One of the Most Important AI Releases of the Year, Tom’s Guide, 7/24/24
Yann LeCun Says Meta AI ‘Quickly Becoming Most Used’ Assistant, VentureBeat, 7/23/24
OpenAI Offers Free GPT-4o Mini Fine-Tuning to Counter Meta’s Llama 3.1 Release, VentureBeat, 7/23/24
Meta’s AI Assistant Is Coming to Quest Headsets in the U.S. and Canada, Engadget, 7/23/24
Meta AI Gets New ‘Imagine Me’ Selfie Feature, TechCrunch, 7/23/24