Meta’s Llama AI engine sees rapid growth in open source adoption

Meta’s Llama AI engine has seen substantial growth in the open source AI landscape, emerging as a leading force in the industry. With nearly 350 million downloads to date, Llama models have experienced a significant increase in adoption over the past year. In July alone, the models were downloaded over 20 million times, positioning Llama as a prominent open source model family.

Since the introduction of Llama 3.1, which expanded context length to 128K and added support for multiple languages, usage by token volume across major cloud service providers has more than doubled in three months. This growth reflects an increasing preference for Llama within the developer community, making it a notable competitor in the AI field.

The open source approach has been a key factor in Llama's widespread use, enabling a diverse range of companies to build AI solutions on-premises or via cloud services. Meta’s commitment to this open approach has facilitated the creation of a dynamic AI ecosystem.

Industry leaders have acknowledged Llama’s influence on AI development. Swami Sivasubramanian, VP of AI and Data at AWS, noted the significant uptake of Llama 3.1 among customers, who are leveraging it for various complex applications. Ali Ghodsi, CEO of Databricks, described Llama 3.1 as a model that effectively combines open source flexibility with commercial-grade quality.

Various companies, including Accenture, AT&T, DoorDash, and Goldman Sachs, are using Llama to enhance productivity and streamline operations. Accenture, for instance, has implemented Llama 3.1 to develop a custom AI model for ESG reporting, achieving notable productivity gains and improved output quality.

As Llama continues to grow, so does its community, with over 60,000 derivative models on platforms like Hugging Face. This open source model has become a preferred tool for developers, who are using it for a range of specialized tasks.