Q* Is Old News Meet New V* AI Which Is on Par with Human Vision (digialps.com)

The dumb misleading SEO clickbait title of 'Q' (which has nothing to do with this) aside, this would benefit from a fairer comparison with GPT-4-V, which is capable of doing various kinds of iteration or selection, like this V does: set-of-mark prompting in particular comes to mind https://arxiv.org/abs/2310.11441