The models themselves support up to 1M. You are just charged more for all context over 400k. The 272k limit is just a client limitation to ensure you can never go over 400k since the maximum output size of the model is 128k.
Not true. This is the case for the API, not for subscriptions: they are hard capped at 400k.
Perhaps not this for this parameter, but for others? Client side.
Apart from the local scaffolding you can configure very few parameters, and the ones you can configure are the ones that don’t matter