SupportedAI toolsKnow which team is behind your OpenAI bill
Split your OpenAI spend by project, team and person. See every model and product, set budgets, and hear about spikes before the invoice lands.
Highlights
Every line item
RootSpend reads every cost line for every project, then splits it by person wherever OpenAI's usage data shows who made the calls.
Ready on day one
Projects and members come in with the first sync. Put people in teams and the costs follow them.
One line for AI
OpenAI, Claude and your seat plans land in the same AI category, so "what does AI cost us?" has one answer.
What you get from OpenAI API
- All models and products: chat, embeddings, images, audio, realtime and fine tuning
- Tools and storage: vector stores, file search, web search and code interpreter
- Cost per person for chat, embeddings, images, audio and moderation
- Projects and members, imported for you
- Budgets, forecasts and spike alerts
What it doesn't do yet
- ChatGPT plans. Add a ChatGPT seat plan for those.
- Azure OpenAI. That spend is on your Azure bill, so connect Azure for it.
- Vector stores, code interpreter and search per person. OpenAI reports those per project only.
The invoice says how much. It doesn't say who.
OpenAI bills your organization one number. Behind it are projects, API keys and people across several teams, all calling different models. When the bill jumps 40%, the invoice can't tell you where the jump came from.
What RootSpend reads
You give RootSpend an Admin API key. It reads four things, all read only.
| Source | What it's used for |
|---|---|
| Costs endpoint | Every cost line, per project |
| Usage endpoint | Tokens and requests by model, project, API key and user |
| Projects | Things you can assign to a team |
| Members | People, so their usage rolls up to their team |
Each cost line is split across the people who used that model in that project, then rolled up to their teams.
2 min
To connect
90 days
Of history on the first sync
4
Read only permissions
What you can do with it
Give each team a monthly AI budget and get an email or a Slack message when the forecast runs over. See how a spike breaks down by project and model. Put OpenAI next to Claude, your cloud bills and your seat plans, and you have your whole AI bill in one place.
Want the long version? Read how to track OpenAI costs per team, project and person.
OpenAI API questions
An organization Admin API key. An organization owner creates it on platform.openai.com under Settings, then Admin keys. A normal project key can't read cost or usage data.
No. The Admin API only exposes billing, usage counts, projects and members. Prompts and completions never reach RootSpend.
A cost line is split by each person's tokens on that model in that project, on that day. Lines OpenAI only reports per project, like vector stores, stay with the project.
The first sync pulls the last 90 days, so you see trends and month over month changes right away.
Want to see your OpenAI API spend split by team?
RootSpend is opening to a first group of teams soon. Join the waitlist, or book a demo and we will show you on sample data.
Opening soon. The waitlist gets in first.