이미지: @OpenAIDevs (X) 화면 갈무리
Summary
- OpenAI has released a feature that lets developer accounts break down usage and spend by individual API key in the Usage/Spend dashboard
- Organizations or projects can now set monthly spend limits, with hard limits available that cut off traffic once the cap is reached
- In a sample screenshot shown by OpenAI, the production-api key accounted for the largest share of total spend — $3,657 out of $5,230 — between August 1 and 20
- 발표 채널
- OpenAI Developers 공식 X 계정
- 발표일
- 2026-08-21
- 신규 기능
- API 키별 Usage·Spend 필터, 조직/프로젝트 월별 지출 한도(하드 리밋 포함)
- 예시 데이터 총 지출
- 5,230달러 (8월 1~20일, production-api 3,657달러·search-indexer 1,018달러·staging 555달러)
OpenAI announced via its official X account that it has added an API-key-level filter to the Usage/Spend dashboard used by developers who work with its API. Previously, the dashboard only showed aggregate spending across an entire organization or project, but now developers can remove the "All API keys" filter and drill down into usage and spend trends for a single key at a time. In the sample screenshot released alongside the announcement, of the $5,230 total spent between August 1 and 20, the production-api key accounted for $3,657, the search-indexer key for $1,018, and the staging key for $555. This shows how wide a cost gap can open up between a key used by a production server and one used for testing, even within the same organization.
An API is essentially a channel through which programs call on one another. Take OpenAI's API as an example: a company running multiple apps — a chatbot, a search indexer, an internal test server — typically issues a separate API key for each one. The problem is that when these keys are grouped under a single organization account, the invoice only shows a total, and developers have had to dig through logs themselves to figure out which app is spending how much. As services scale, this tracking burden grows. In fact, in early August, AWS unveiled an architecture that automatically collects usage data for OpenAI Codex running on Amazon Bedrock through CloudWatch and breaks it down by team, department, and cost center. OpenAI's new feature can be seen as addressing the same underlying issue: that organizations often struggle to see for themselves who is spending how much.
With this update, developers can select a specific API key in the dashboard to view daily spend trends and quickly identify apps or workloads that are costing more than expected. They can also set monthly spend limits in advance at the organization or individual project level, with hard limits available that block traffic entirely once the cap is hit. Organizations where multiple teams share a single OpenAI account can use this by issuing separate keys per team and then using the dashboard to prevent budget overruns before they happen.
Comments