Tencent's Youtu Lab published a paper measuring and comparing the visual understanding of Google's Gemini Pro with OpenAl's GPT-4V, and also comparing it with the open-source MLLM Srhinx, which shows the disparity between artificial and black-box systems.
