With all the recent and future changes to the pricing, wouldn’t it be handy to show the user how much a model costs right in the model selection popup?
Hell, maybe we can even figure out like predicted token usage in comparison to some “golden standard” model as some models like to generate a lot of bs and / or think entire books on one tool call. And then we can predict something like “nominal” price for a request!
So, my proposition is to at least give us clear pricing info for each model right in the UI.
Then, if it’s realistic and viable, to implement some sort of a metric of how much tokens each model usually generates on some nominal task and from that calculate its “request price” (which obviously won’t reliably predict the real usage but at least will give us a perspective of how much will it cost compared to other models.