Your rate with each AI vendor

What each AI provider really costs as a share of its list price, whether the price you agreed is on the bill, and what each seat tool costs per person who used a seat.

Vendors › Rates answers two questions a bill can't: what you really pay each AI vendor, and whether the discount you negotiated is on the bill.

How your rate is worked out

Your rate is what a provider billed, divided by what the same usage costs at its public list price. A 20% discount makes it 0.80×; paying list makes it 1.00×.

1. Veqen reads the provider's own usage report, per key, hour and model (Protection connects it).
2. When each hour is read, Veqen prices it twice: at the list price in the price book and at your prices. Each hour is priced as that many average calls, with the call's tier (priority or batch), cache reads and writes and long-context rates, so a busy hour never lands in a tier no single call reached.
3. The bill comes from the provider's cost import (Bill check). Azure, AWS Bedrock and Google Cloud bills carry token quantities line by line ("…input tokens", "…Outp regnl Tokens"), so Veqen values each standard token line at list and at your prices when it imports the bill. Lines priced apart from a model's standard rates (batch, priority, regional and data-zone deployments, long context, audio, images, storage, search) stay out of the rate, and Google's credits don't count as a discount.
4. Only the days both the bill and the usage cover count. Charges that aren't tokens, such as web search or storage, and billed models with no usage on those days are shown apart, not in the rate.

A provider whose usage Veqen doesn't read is listed with its bill and marked not checked: part of its usage could be missing, and that would look like an overcharge. OpenAI's and Anthropic's own usage is read with protection on a paid plan, or during the free audit.

Billed above what you agreed

When your prices are on file (a provider-wide discount, or your own rate per model), Veqen compares the bill with the same usage at those prices:

The ruler on each row shows list price, the rate you agreed and the rate you were billed, so a dot to the right of the agreed mark is money above your contract.

Seat tools

For each coding tool with a seat price in Tool contracts, Rates shows the seats you pay for, how many were used in the last 30 days, and what each person who used one really costs: seats nobody used push it above the seat price. Release them from Seats.

Who sees it

Owners, admins and finance members, the same people who see bills. Members, viewers and security members don't.

Over the API and MCP

• GET /v1/rates?start=YYYY-MM-DD&end=YYYY-MM-DD with a full-access key; the last 30 days by default.
• The MCP tool vendors with view: "rates".

Both return each provider's billed_usd, list_usd, agreed_usd, rate, agreed_rate, above_agreed_usd, state (act, watch, clear or off), checked, covered_days, its models, and each seat tool's seats, used, per_used_usd and idle_usd_month.