Show HN: Self-hosted AI spend attribution (demo.costura.ai)

[flagged]

How do you keep track of changing model costs, new models, etc? Do you need to update it manually each time?

I had the same issue with Anthropic not reporting cost per key. I ended up adding an “estimated” figure too, so it could at least be compared with other services. Do you label the split as an estimate in the UI, or show it like the other providers’ numbers?