Not showing the uncached token count prominently makes the total token usage seem much higher than it is as cached tokens are very cheap.
Gemini 3.0 makes a large # of small requests that much better utilize cached tokens which makes the total token count very misleading.
We should instead show the number of non-cached tokens as the input tokens as that is the meaningful stat for most users.
We should likely indicate the # of cached tokens at least for API Key users but need to make sure we don't do it in a way that causes people to be confused about cost of overall efficiency.