Not a fan of Meta generally, but they’ve been leading in open-source AI. And yes, I understand how counterintuitive that seems - perhaps Zuck is chasing his 'redemption arc'.

They released LLaMA over a year ago, which is their open-source family of large language models - considered to be best-in-class for research, efficiency, and accessibility, compared to other open-source models like GPT-3 and even proprietary models like GPT-4 [which was the leading model up until the more recent 4o + o1].