We want to make it more transparent which of our features consume the most tokens. I therefore found the Custom Reporting API which sounds perfect. We could use the users/tags to apply vertical/feature based tagging to perfectly see a breakdown.
But: We would like to use the regular AI Gateway Overview at https://vercel.com/.../~/ai-gateway?period=30d to display these metrics. It already has the ability to display tokens/cost by provider/model and down below you can get the request summary by project or API key. This would be a perfect spot to also display a summary grouped by “user” or “tags”. Is this planned? Since we would like to avoid to build our own custom internal UI just to display an AI usage summary.
Right now, we are going to use a bunch of API keys to separate API calls since this gives a perfect overview of which API key (in our case which feature) consumes how many tokens/cost.