Track OpenAI and Anthropic Spend by Team With Concentrate

On this page

A shared provider key can tell you which account was billed without telling you which team created the cost. Concentrate gives teams a shared way to buy model inference and inspect spend by organization, team, project, key, model and provider. Useful attribution starts with assigning requests to the right owner.

Give each workload an identifiable key

Separate customer-facing traffic, internal experiments and coding agents into appropriate projects and keys. Assign an owner to each. Avoid distributing one application key across unrelated teams if you need their spending separated later.