Meta releases new version of its AI model – and makes it open to the public
Meta has released Llama 3, a new AI model that it claims is the state of the art.
As with much of its previous work in artificial intelligence, Meta will open source the new model, allowing the public to examine and use it. That puts it in competition with Google and OpenAI, which have refused to make most of their work on similar systems open to the public.
Meta has argued that making the system widely available will help encourage innovation and allow people to use AI in a variety of ways. It can be downloaded through a devoted website.
Llama 3 is a large language model, of the same kind that powers OpenAI’s ChatGPT and Google’s Gemini. It was built with the aim of making a system that was “on par with the best proprietary models available today”.
“We wanted to address developer feedback to increase the overall helpfulness of Llama 3 and are doing so while continuing to play a leading role on responsible use and deployment of LLMs,” it said. “We are embracing the open source ethos of releasing early and often to enable the community to get access to these models while they are still in development.”
The version being released today is just the first in Meta’s collection, it said. Future versions should get multiple languages, make it multimodal so that it can use a variety of kinds of content, and add better performance.
Meta said that the new model is now being used in Meta AI, its chatbot, which can be used in Facebook, Instagram, WhatsApp, Messenger and on the web, though is not available in some countries including the UK. That system is now the most intelligent system you can use for free, it claimed, suggesting that it believes that it is better than the base version of ChatGPT.
Meta AI has run into some controversy in recent days after it claimed to have a “child who is both gifted and has a disability”, as it attempted to help with questions in a group of parents.