Meta should be the name on everyone’s lips in the coming weeks when we talk about AI. Because the new version of its Llama language model, Llama 3, is about to be made public.
Mark Zuckerberg’s firm does not want to leave the AI sector to its rivals Google (with Gemini) and Microsoft (Copilot). It has been developing its own language model for more than a year, which is currently at the level of the Llama 2 version. And this will very soon see an update, with Llama 3, which Meta wants to update very quickly. on the market. And this, in several stages.
A preview next week for Llama 3…
Llama 3 is not far away! The language model created by Meta is indeed ready for distribution. And if we are to believe the information from The Informationnext week we will be able to get our hands on two “small versions” of Llama 3, which will be used, in particular, to tease the full version.
“ There will be a number of different models with different capabilities, different versatility over the course of this year, starting really soon » summarized Nick Clegg, President of International Affairs.
…before a release next month
But what is most impatiently awaited is obviously the full version of Llama 3. And even if the Meta teams did not want to communicate a precise date for its release, the tech giant has everything also indicated that it should normally be made public during the month of May.
According to information communicated this Wednesday, April 10 during Meta’s Innovation Day, Llama 3 will, like its competitors, be multimodal. It will also very quickly be integrated into the entire Meta ecosystem. Will it have the same success as the Gemini and ChatGPT AIs?
Source : Medium, TechCrunch, Tom’s Guide
2