Costs
Open core · self-host + all Splyntra Cloud plans
Cost analytics is on every plan. Splyntra turns the token counts on your llm_call
spans into dollar spend and rolls it up so you can see where the money goes.
Spend attribution
The Costs screen breaks spend down three ways:
- By run — the cost of a single agent run, shown alongside its trace.
- By model — which models (and providers) dominate your bill.
- By project — spend rolled up per project (and filterable by environment).
The table is searchable, sortable, and paginated, with Excel export for reporting.
Manage model pricing
Cost is tokens × price, so pricing must be right. Use Manage model pricing to view
and edit the per-model input/output token prices Splyntra uses. Add a model that isn't
priced yet — common when you call an OpenAI-compatible provider
whose models aren't in the default table — and its calls become cost-attributed
immediately.
Budgets & cost alerts
Set budgets to track spend against a target, and pair them with cost alerts so a run of expensive calls doesn't surprise you at month end. Cost-overrun alerts are one of the two alert rule types; see Alerts for channels and rule setup.
Next steps
- Metrics — spend over time alongside latency and error rate.
- Alerts — create a cost-overrun rule.
- Cost analytics — how token-to-cost attribution works.