Deepmind creates these, Google Cloud productizes them.
If Google Cloud was in charge, Gemini 5 would be public, and they wouldn't be messing with Gemini 4 nor 1M context when they have 10M internal. C'est la vie.
A 10M window would require dozens of GPUs (or TPUs) dedicated to each query. That is something that Google could afford internally but wouldn't be able to scale to an external product.
If Google Cloud was in charge, Gemini 5 would be public, and they wouldn't be messing with Gemini 4 nor 1M context when they have 10M internal. C'est la vie.