Text is a compression and abstraction of the world, while vision is the world’s own projection.
Most people chose the former, like OpenAI; Google chose the latter, betting from the very beginning of the Gemini series on the “native multimodal” technical path and entering the “narrow gate.”
By the time Gemini 3 arrived, Google had combined native multimodality with deep reasoning ability in an unprecedented way, and only then did the whole market see the result of that choice of path.