Observability
Reading the cost numbers
Most AI is charged by tokens: roughly, pieces of words. In is what the AI reads (your prompt, documents, context); Out is what it writes. Longer source material and longer answers cost more.
Why two runs of one workflow cost different amounts
- How much it read. A step grounded on a project reads its documents: the more it reads, the more tokens in.
- How much it wrote. A long summary costs more than a short one.
- How many steps ran. A run that stops early costs less than one that finishes.
How it relates to your credit
The spend on Observability is the same spend that draws down the account's credit. Billing shows the balance and lets you add credit; Observability shows where it went. Billing and adding credit
A number looks wrong? Start with AI spend by feature, then By workflow. A surprise is usually a scheduled workflow running more often than intended, or one grounded on far more of a project than it needs.
Signed in to VeraGen? Press Guide me on any screen for a walkthrough, or ask Odin.