History
- Jul 30, 2026Compute usage is now metered by flavor.
Usage, attributed
Tokens, compute, and storage — rolled up per agent, per team, per model.
WhereConsole
Introduced Apr 12, 2026 · updated Jul 30, 2026
Tokens
84.2M
Cached
61%
Compute
42h
Spend
$612
rolled up per agent · per team · per model
How it works
Cost lands on who drove it.
Tokens, compute, and storage roll up per agent, per team, per model — a fleet becomes chargeable to the teams that run it, not one line item nobody owns.
The provider's own tokenizer.
Tokens are counted the way the provider counts them, so the number in the report is the number on the invoice.
Cache, broken out.
Cached input is counted separately — cache hit ratio is the leading indicator of what a fleet actually costs.
See it on your own cluster.
Thirty minutes, your infrastructure, your stack. Or skip the call — it is one Helm release onto a cluster you already run.