Back in April, the company said it would move all Copilot plans to a usage-based system that bills users based on actual AI consumption, measured in tokens,...
I think that was before. Now it’s based on the compute cost of the request. Like if you ask a large model to perform a long and difficult task you’ll pay through the nose. If you ask a simple model to perform an easy task it’ll be dirt cheap. And that’s what’s got people pissed, is that what used to be ‘one query’ now may consume a lot of credits and it may not be clear how many.
what is a “usage” btw? They said 1 cent = 1 usage.
So 1 message sent to Copilot? Or usage = number of response Copilot gives back to the user?
I think that was before. Now it’s based on the compute cost of the request. Like if you ask a large model to perform a long and difficult task you’ll pay through the nose. If you ask a simple model to perform an easy task it’ll be dirt cheap. And that’s what’s got people pissed, is that what used to be ‘one query’ now may consume a lot of credits and it may not be clear how many.