See what your company spends on OpenAI, Anthropic, Google, AWS, Azure and every other model provider — by team, by project, by model. Set budgets, get alerted early, and give finance numbers they can trust.
Why teams switch from spreadsheets
Each provider has its own console, its own invoice and its own idea of a project. AI Cost Control puts them side by side, normalised to dollars per day.
Model APIs, clouds and routers in a single daily ledger, with the same model and project vocabulary everywhere.
Map OpenAI projects, Anthropic workspaces, Azure resource groups or GCP projects to the teams that own them.
Monthly budgets for the company, a provider or a team. Alerts at the thresholds you choose, by email or Slack.
Compare periods, spot the day a new feature shipped, and see which model drove the jump.
One CSV with day, provider, model, project and team. Drops straight into your FP&A model.
One alert per threshold per month. No noise, no dashboards nobody opens.
Providers
Automatic sync wherever a provider offers a billing API. A guided CSV import everywhere else.
…and any other provider whose usage you can export.
How it works
Paste a read-only billing credential for each provider. We check it works before saving it.
We list every project and workspace we find. Name them after your teams once.
Watch daily spend, set monthly budgets and let alerts tell you when to look.
Security
Cost data is sensitive and credentials even more so. Here is exactly what we do with both.
We read billing reports only. Prompts, completions and end-user data stay between you and your providers.
Read-only cost permissions where providers allow it: ce:GetCostAndUsage on AWS, Cost Management Reader on Azure, BigQuery Data Viewer on Google Cloud.
Encrypted before storage (AES-128 + HMAC-SHA256), never shown again, never sent to your browser.
Every query is scoped to your organisation. Isolation between workspaces is covered by automated tests.
Argon2 password hashing, revocable server-side sessions in HttpOnly cookies, rate-limited sign-in, HTTPS everywhere with HSTS.
Export everything as CSV at any time. Removing a connection deletes its credentials and its cost history.
FAQ
No. We read billing and usage reports from each provider's admin API. Your application traffic never goes through us, so prompts and completions never reach our servers.
We ask for admin or billing credentials that can read cost reports. Where the provider supports it (AWS, Azure, Google Cloud), you grant a single read-only cost permission. We never call model endpoints with your keys.
Each credential is encrypted before it is written to the database (Fernet: AES-128 with an HMAC-SHA256 signature) and is never displayed again. Delete a connection and its credentials and cost history are deleted with it.
Mistral, xAI, DeepSeek, Groq and others don't publish a cost API yet. Upload their usage export as CSV and it lands in the same dashboard, budgets and reports. Re-uploading a month replaces it, so there are no duplicates.
We sync every few hours. Providers themselves publish cost data with a delay (a few hours for OpenAI and Anthropic, up to a day for the clouds), so the dashboard reflects what your invoices will show.
Budgets are alerts today: email and Slack when spend crosses 50%, 80% or 100% (thresholds are yours to set). They never interrupt your production traffic.
Connect your first provider in minutes. Up to 90 days of history are imported on the first sync.