I think Flamingo might be a bit better, but I continue to be amazed at the accelerating pace of AI models, especially LLMs, these days. And they all seem to incorporate transformers, too, which makes me very much appreciate Vashwani, et al's 2017 invention of them. There could be a better model out there to discover, but so far transformers truly feel like the key we've been looking for to unlock AGI, or at least something very close to it.