I kept checking the OpenAI dashboard all day… so I asked ChatGPT to fix it
When you're building with OpenAI / Claude APIs, the annoying part isn’t the models.
It’s token costs.
You ship a feature, run prompts, maybe test an agent…
Then suddenly your usage jumps and you have no idea what caused it.
So I kept doing this all day:
Open dashboard → check usage
Close it → go back to coding
Open dashboard again 20 minutes later
Eventually I got tired of that and asked ChatGPT if there was a way to see token usage while I work instead of constantly opening dashboards.
That led me to Tokenbar.
It’s a small macOS menu bar app that shows live token usage and estimated cost while you’re working.
So instead of checking dashboards all the time, you can just glance at the menu bar and see if something is burning tokens.
I’ve been using it while building and it’s actually pretty helpful.
Curious how other people here track LLM usage right now.
Do you just check the provider dashboards or do you have something monitoring it?
