How do I tell traffic growth apart from a broken agent in my LLM bill?

Look at dollars per turn, not total spend. More users running more turns and a broken agent both drive total cost up, and they only separate once you divide: growth adds turns at roughly the same price each and leaves cost per turn flat, while a defect, a prompt that grew, a retry loop, output that got wordier, moves the per-turn number on its own, whether traffic changes or not.

That ratio is what cost_drift already tracks, per call site, against that call site’s own recent history rather than a company-wide total. A busy call site and a rarely used one each get judged against what normal looks like for them, so a spike in one doesn’t need a spike everywhere else to register.

Total spend still matters for the finance conversation. It’s just the wrong number to stare at when you’re trying to find a bug.

keep reading

More on this.

Two ways to run Tessary.

Tessary is an open-source agent reliability platform. Cloud and self-hosted run the same workflow on the OpenTelemetry traces your agent already emits.

Tessary Cloud

We host it for you. Send your first trace with nothing to deploy and no model key.

what's includedper organization
traces
10,000 per calendar month
stored trace data
1 GB
retention
30 days
model credit
$10, one-time, for triage and root-cause analysis
credit card
not required

Self-hosted Tessary

Run the open-source code on your own infrastructure with one command. Add your own model key for triage and root-cause analysis.

Self-host Tessary for me by following https://github.com/tessaryai/tessary/blob/main/setup.md

docker compose -f oci://docker.io/tessaryai/tessary:compose up -d -y