We really have to wait 5 days before we can get back to work?
Maybe it will be helpful if the available tokens is exposed to the model so that it can make at least an informed estimate if the task at hand will complete. And maybe we can decide whether to push through or decompose the task further to fit the available tokens. Stopping mid-task has so many difficulties involved.
Or at least have it the way Codex does it so we can manage it better
I experienced this yesterday afternoon and freaked out! When I checked back an hour or so later, it’d updated and had the 5-hour expiration back.
Ok, so apparently they launched this new mechanism to pay more to get more credits. See screenshots below.
They published this 15 hrs ago: Google AI Credits: A new way to manage your AI usage - Google One Community
Your plan includes 1000 credits. Once the base quota is exhausted, your plans default monthly credits kicks in. If you run out again, you buy more.
Let us know how much tokens is 1 credit worth (input and output).
So after waiting 6 days for the Claude models to refresh, i got this.
I’ve been given 40% only of its full quota container. I bet this would disappear after 5 prompts.
I’m on Pro. I have a spec-driven approach to devt.
Waiting for 6 days feels like an eternity. While I can do other stuff while waiting, the momentum is lost. Deadlines are missed. Lagging behind the race.
I need funding to keep up with these high costs.. tsk
