The big change about this announcement is the 1M context window on all models.
But the price is what matters.
Nothing compared to Llama 4's 7M. What matters is how well it performs with such long context, not what the technical maximum is.