Usage & Logs

Every run is logged. The Usage & Logs dashboard breaks down cost, latency, and error rate by model, prompt, and version.

Where to find it

Open Settings → Usage & Logs. Every UI run, compare, optimizer call, and API call is recorded best-effort to your run history, capturing the provider, model, token counts, estimated cost, latency, and status.

Three views

TabShows
OverviewSummary cards, a cost chart over time, a by-model table, and your most recent runs.
By PromptA sortable table of your prompts by run count, with average latency and error rate.
By MemberOwner-only, Team plan: a per-teammate breakdown of run count and cost.

Per-prompt analytics

Individual prompts and chains also expose per-version run stats, so you can see how a specific version is performing in production right alongside its evals.

Optimizer spend is included
Because optimizer calls are logged with source="optimizer", the cost of analyzing and improving prompts shows up here too — not just your production traffic.

Budget alerts

Set a monthly spending cap in Settings → Account → Budget. PromptVersions emails a warning at 80% of your budget and blocks further runs once the limit is reached. The budget resets at the start of each calendar month.