New model benchmarks GPT-4, to close performance gap, Meta is accelerating AI development

In an effort to close the gap with competitors like $Alphabet-A(GOOGL.US)$ , $Microsoft(MSFT.US)$ , and OpenAI, $Meta Platforms(META.US)$ will begin training a new model next year, possibly open-sourcing it.
According to Microsoft sources familiar with the matter, the new AI model under development, which is benchmarked against the GPT-4 introduced by OpenAI, could help the company build services that can generate complex text, analytics, and other outputs.
To build the data centers needed to work on the project, Meta is acquiring more H100s, and while Meta has partnered with Microsoft to make Llama 2 available on Azure, Microsoft's cloud computing platform, it plans to train the new model on its own infrastructure, according to people familiar with the matter.