Meta Releases 3.1 Llama, New Open Source AI Model Claimed To Be More Sophisticated

Meta CEO Mark Zuckerberg (photo: Meta)

JAKARTA - Meta released a new artificial intelligence (AI) model, which is claimed to be able to compete with technology from OpenAI and Google, and make it free for anyone to use.

Called Llama 3.1, Meta is now expanding its strategy by providing an open source AI model, which means anyone can use or modify the model.

I believe that AI will develop in the same way. Currently, several technology companies are developing a prominent closed model. However, open sources are rapidly closing those gaps,' said Meta CEO Mark Zuckerberg in his latest blog.

Meta released the final version of its AI model, Llama 3 a few months ago. With this new open source AI model, Meta hopes that Llama can become the most advanced model in the industry.

"Starting next year, we hope that future Llama models will be the most advanced in the industry. However, even before that, Llama has led in terms of openness, modifyability, and cost efficiency," he added.

In addition to having a much better cost/performance compared to a closed model, Zuck said that this open source model would make it the best choice to fine-tune and screen smaller models.

In addition to releasing these models, Meta is also working with a number of companies to develop a wider ecosystem such as Amazon, Databricks, and NVIDIA to help developers implement their own versions.

These models will be available in all major clouds including AWS, Azure, Google, Oracle, and more. Companies such as Scale.AI, Dell, Deloitte, and others are also said to be ready to help companies adopt Llama and train specific models with their own data.

Meta also claims that the development of Llama 3.1 costs about half of OpenAI's GPT-4o costs to run it in the production stage.

Meta is committed to open source AI. I will outline why I believe open sources are the best development stack for you, "said Zuck.