CEO of the company Mark Zuckerberg made the announcement in a video on Instagram.
“We have another big AI release today,” Zuckerberg said in the video. He said that “Meta is on track to become the most used AI assistant in the world by the end of the year,” adding that hundreds of millions of people use it every day. The tech company CEO also said that the AI model will soon be available in more countries.
How is Llama 3.1 different from Llama 3?
Zuckerberg added that Llama 3.1 will support more languages, offer more reasoning, and be “overall better” than its predecessor. He noted that the company is adding a feature that allows it to generate images that “users can put themselves in.”
According to the company, the new model is released with three training parameters: Llama 3.1 with 405 billion parameters, the first frontier-grade open-source AI model, as well as the improved Llama 3.1 models with 70 billion and 8 billion parameters.
“In addition to having a significantly better cost-performance ratio compared to closed models, the fact that the 405B is open will make it the best choice for refining and distilling smaller models,” the company said. Zucekrberg also noted that Meta will also launch a tool that will allow customers to create their own AI models.
Meta works with Amazon and Nvidia for developer access
Similar to Meta’s release of Llama 2, the company is partnering with other technology companies that will offer their customers access to Llama 3.1 through their respective cloud computing platforms.
“Beyond publishing these models, we are working with a range of companies to grow the broader ecosystem. Amazon, Databricks and Nvidia “We are launching complete suites of services to help developers refine and distill their own models,” Meta added.
The largest version of Llama 3.1 (with 405 billion parameters) was trained with over 16,000 Nvidia H100 GPUs.
