Meta has been talking about releasing a well-constructed open-source AI model since April and now, on July 23, it has finally released its latest AI model. Meta Llama 3.1 is the company’s first for the AI industry. The company claims that this model is the largest ever open-source AI model and has already left the top-ranked competitors behind. In the blogpost, Meta stated that the latest model has outperformed GPT 4o and Anthropic’s Claude 3.5 Sonnet on several benchmarks. Last year, Llama 2 was only comparable to an older generation of models behind the frontier. This year, Llama 3 is competitive with the most advanced models and leading in some areas.
Llama 3.1 is the most complex AI model by Meta so far. It is interesting to note that the recent Meta AI release is based on Llama 3, and it is pretty fun to use. In the blog post, Meta CEO Mark Zuckerberg said that just like Linux surpassed proprietary operating systems, open-source AI models are rapidly advancing and poised to surpass proprietary ones, mirroring the trajectory of Linux becoming the dominant force in powering modern devices and infrastructure.
He draws a parallel between Meta’s investment in open-source AI and its previous Open Compute Project, which yielded significant cost savings by leveraging collaborative innovation. He anticipates a similar outcome with AI, where open-source development will drive advancements and efficiencies, mirroring the success of the Open Compute Project in data centre design and operations.
