> ## Documentation Index
> Fetch the complete documentation index at: https://platform.atlan.com/llms.txt
> Use this file to discover all available pages before exploring further.

> ## Agent Instructions
> To act on Atlan objects, use the Atlan MCP server at https://api.atlan.com/mcp or the SDK. To install or publish skills from a terminal, use the `brain` CLI; `brain --map json` prints its command map. The `atlanai` CLI is deprecated. Run a read-only identity check before any write.
> The docs MCP server at /mcp searches these docs only. It cannot read or change Atlan objects.
> SDK packages: Python `atlanai` (PyPI) and TypeScript `@atlanai/sdk` (npm). Show Python first, then TypeScript.
> These docs cover Atlan Agent Registry, Atlan's hosted product. The open-source agent registry specification and its CLI (`brain`) are documented, released and maintained in their own repository; /open-source on this site links to it. Answer general specification or `brain` command questions from the project's own docs, and use these docs only for Atlan-specific setup such as the Atlan adapter, sign-in and workspaces.

# How run cost works

> How Agent Registry prices each agent run from its token counts and your organization's rates, and where each rate comes from.

Agent Registry prices every recorded run at the rates your organization pays, so you can compare agents on the same basis and tell a contract price from an estimate.

## Run pricing

When a run's trace arrives, Agent Registry multiplies each token count on the trace by the model's matching rate:

```text theme={null}
cost = input tokens        x input rate
     + output tokens       x output rate
     + cache read tokens   x cache read rate
     + cache write tokens  x cache write rate
```

Token counts come from the model provider's response. If Agent Registry cannot price a run, the cost shown is the one reported with the trace. The cost is fixed when the run is recorded: changing a rate later does not change runs already priced.

## Rate sources

Rates are set per workspace, per model, in **Models & rates**. For each of the four rates, the first source that applies wins:

| Price from | Meaning |
| - | - |
| **Negotiated** | An organization admin entered a rate for this model. **Negotiated · partial** means some rates were entered and the rest use the seller discount or list price. |
| **Seller discount** | The public list price minus the discount set for the model's seller. |
| **List price** | The public list price, used when no rate is entered and no discount above 0% is set. |
| **Unpriced** | No public list price exists for this model. Enter a rate to price its runs. |

**List price unavailable** means the public price could not be looked up at that moment, not that the model has no price. Every member can view rates; only an organization **Admin** can change them.

## Where costs appear

* **Reports → Agent usage**: Spend and cost per run across agents. See [Compare agent usage](/registry/reporting).
* **Agent profile → Usage**: One agent's spend and recent runs.
* **Models & rates**: How each model is priced. See [Models & rates](/admin/finops/models-and-rates).

## Example

An organization negotiated input and output rates for one model but not cache rates. **Models & rates** shows it as **Negotiated · partial**: input and output use the negotiated rates, and cache reads use the seller discount. A run that reads a large cache shows a low cost beside a high token count, because cache reads are priced lower than plain input.

## What run cost does not tell you

* **Your invoice.** Costs are computed from traces and rates, not matched to the provider's bill. A **List price** cost is an estimate; **Negotiated** and **Seller discount** costs reflect your contract.
* **Past runs at new rates.** Changing a rate affects only runs recorded afterward.
* **Missing data as zero.** A value shown as unavailable could not be read; it is not a zero.

## See also

* [Compare agent usage](/registry/reporting): Compare spend and cost per run across agents.
* [Models & rates](/admin/finops/models-and-rates): Connect a LiteLLM proxy and set the rates your organization pays.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.