Billing & usage
Understand where usage goes, drill into individual calls, and request read-only analysis with a clear credit limit.
Spend includes credits charged, the estimated list value of bring-your-own-key usage, and free promotional usage at list price. Hover or focus a figure for its components. BYOK estimates and free list value are not debits from your credit balance.
Amounts are kept in integer nano-dollars: 1,000,000,000 equals one US dollar. Report APIs carry amounts as decimal strings and CSV dollar columns retain nine decimal places. Cached input is part of input tokens; reasoning is part of output tokens, not extra volume.
Choose a preset or custom dates in the date dropdown, select your timezone, and group by model, API key, recorded member, end-user label, provider, funding, API surface, outcome, error, repeated prompt, usage source, or a tag key. Multiple values within one filter mean OR; separate filters mean AND.
API key ownership is not proof of the human behind every request. End-user labels are supplied by the caller. Missing attribution is kept visible rather than guessed. Click a group to narrow the report, or choose View Activity to inspect the same exact filters and dates.
Export CSV includes every group in the selected snapshot, not just the visible page. Export supports up to 10,000 groups. Very large windows fail explicitly; narrow the range instead of interpreting a partial result as a complete total.
Spend reports settlement-time usage. Historical projection can take time; incomplete coverage and unknown provider usage are labeled. Accepted failures before provider dispatch count as requests. Protocol rejections before admission are currently outside this projection.
Application headers are not yet a collected reporting dimension. Tag grouping reads captured tags only; older calls remain untagged. The shared tag header requires the corresponding gateway engine release and is not available merely because the reporting filter exists.
Intelligence is Pro-only and read-only. Ask a question about cache usage, errors, or spend contributors, or select Build me a report. It uses GPT-5.6 Luna and sends your question plus content-free aggregate evidence to the model, never stored prompts or completions.
Set a maximum before starting. The default is ten credits, a ceiling rather than a query price. Actual model cost is passed through, including fractional credits, with no fixed fee or markup. Unused reserved credit is released when the run settles.
Saved analyses stay inside your organization and include evidence links and limitations. Model alternatives are candidates to evaluate, not automatic configuration changes or guarantees of equal quality. Opening the page never starts paid work.
Never put secrets, prompt content, email addresses, or personal information in reporting labels or tags. Labels are metadata. Privacy erasure removes caller attribution while preserving anonymous financial amounts.
See the Platform privacy policy and Telemetry for data controls and usage APIs.