As competition in the AI space heats up, Meta Platforms (NASDAQ:META) is developing a new AI model aimed to be as powerful as OpenAI’s latest ChatGPT version GPT-4, the Wall Street Journal reported.
The new model, which could be ready next year, is being designed to be a lot more powerful than Llama 2 that was released two months ago.
Meta’s (META) latest AI model will help companies build services that produce text, analysis and other output, the report said citing people familiar with the matter.
Training for the new AI system – developed by a top-level AI product group formed by Meta (META) CEO Mark Zuckerberg – is expected to start in early 2024.
It was earlier reported that Meta’s (META) AI models would be open-source. The latest model will be trained on Meta’s (META) own infrastructure, unlike Llama 2 that was developed in collaboration with Microsoft (MSFT).
The updates come ahead of a summit on Wednesday organized by Senate Majority Leader Chuck Schumer to discuss AI regulation. Several top tech CEOs are expected to attend, including Zuckerberg and OpenAI CEO Sam Altman.