Meta Launches Largest Llama 3 AI Model, Citing Language and Math Advancements

Nitin Konde, Apoorva Bajj

Jul 25, 2024 — 2 min read

Meta Launches Largest Llama 3 AI Model

On Tuesday, Meta Platforms (META.O) unveiled the most recent version of its Llama 3 AI models, which are primarily available for free. These models compete with paid models from competitors like OpenAI and offer superior multilingual skills and general performance metrics.

Announcing the release of the new Llama 3 model, the parent company of Facebook stated in blog posts and a research study that it can talk in eight languages, produce higher-quality computer code, and solve more complicated math problems than earlier versions.

Although it is still smaller than top models supplied by competitors, this new version outshines the previous one with 405 billion parameters, or variables, that the algorithm considers to produce answers to user questions.

During his extensive promotion of Llama 3, Facebook CEO Mark Zuckerberg stated he expects that the future Llama models will surpass proprietary competition by the end of next year. He claimed that hundreds of millions of people were already using the Meta AI chatbot, which was driven by those models, and that it was on course to become the most popular AI assistant by the end of the year.

Competition Is Getting Stiffer

The release aligns with the ongoing competition among technology companies to demonstrate that their expanding portfolios of resource-intensive large language models can generate substantial improvements in well-established problem areas, such as advanced reasoning, to justify the substantial investments they have made in them.

According to Meta's leading AI scientist, these models will eventually run into reasoning limitations, necessitating the development of new AI systems.

Amazon is reportedly working on a model with two trillion parameters, whereas OpenAI's GPT-4 model reportedly has one trillion.

Meta announced that it is updating its flagship 405 billion parameter model as well as its 8 billion and 70 billion parameter Llama 3 models, which debuted in the spring but are reduced in weight, with new versions coming out soon.

Meta has announced that it will be releasing revised versions of its flagship 405 billion parameter model, as well as its 8 billion and 70 billion parameter Llama 3 models, which debuted in the spring. The models are lighter in weight.

With an expanded "context window," the three new models can handle larger user requests and are multilingual.

Benefits Llama offers to its users

Meta makes its Llama models available to developers for little or no cost, an approach that Zuckerberg claims will lead to better products, less reliance on potential rivals, and higher engagement with the company's primary social networks.

If developers choose the company's free models over its paid ones, it would hurt its competitors' business models and benefit the company too. Meta highlighted potential improvements in important mathematics and knowledge exams in its announcement, which may increase the attraction of that possibility.