> ## Documentation Index
> Fetch the complete documentation index at: https://arize-ax.mintlify.site/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Cost Agent

> Breaks down spend by model, task, and time window, finds expensive models doing low-complexity work, and recommends cheaper routing with quantified savings.

<Info>
  The Cost Agent is part of managed agents on **Enterprise**. Through **September 15, 2026**, Signal and managed agents are included at no additional charge for Enterprise. See [Pricing and usage](/docs/ax/security-and-settings/pricing-and-usage).
</Info>

The **Cost Agent** breaks down spend by model, task, and time window, finds expensive models doing low-complexity work, and recommends cheaper routing with quantified savings.

Run as an automation, it files ranked **cost issues** in the same place [Signal](/docs/ax/observe/signal) files its own — scoped to spend rather than correctness. Run as a session, it returns the analysis in the transcript.

## What it finds

| Pattern                                                                    | Quantified as                                                  |
| -------------------------------------------------------------------------- | -------------------------------------------------------------- |
| **Over-provisioned model tier** — a costly model doing low-complexity work | Current spend vs. spend at a cheaper tier, on that call volume |
| **Redundant or duplicate calls**                                           | Duplicate count and summed wasted cost                         |
| **Runaway retries** — repeated calls after a failure in one trace          | Retries per trace and their cost                               |
| **Expensive prompts** — usually unbounded context stuffing                 | Prompt-token percentiles and marginal cost                     |

It clusters by cause, so one finding covers a recurring pattern rather than a single expensive call, and each carries the numbers behind it plus a proposed fix.

<Frame caption="A session run: total spend for the window, a per-model breakdown, and ranked bottlenecks with a costed fix for each.">
  <img src="https://storage.googleapis.com/arize-phoenix-assets/assets/images/arize-docs-images/agents/ax-agent-cost-result.png" alt="Cost agent session transcript reporting a 41 dollar total bill across 3,837 LLM spans over 90 days, a cost-by-model table with per-model share and average cost per call, a weekly spend spike, and a top bottleneck recommending a cheaper judge model with the saving quantified" />
</Frame>

<Note>
  **Every figure comes from your span data; the agent never estimates.** It reads an explicit cost attribute where one exists, otherwise multiplies token counts by the model's published price, and excludes any span with neither — reporting how many it dropped. So accuracy depends on your spans carrying tokens or costs: if token columns show `n/a`, fix instrumentation first with [Track costs](/docs/ax/instrument/track-costs).
</Note>

## Run it

**Signal & Agents → + New Agent → Cost Agent.**

<Frame caption="Review & launch: the harness, the project it will read, and the run mode, with the task prompt editable before you create the automation.">
  <img src="https://storage.googleapis.com/arize-phoenix-assets/assets/images/arize-docs-images/agents/ax-agents-cost.png" alt="Cost Agent setup on the Review and launch step, showing Claude Code as the agent, a bound tracing project, Automation mode on a schedule, and the editable task prompt describing cost breakdown by model and task type" />
</Frame>

| Setting          | Default                                                                      |
| ---------------- | ---------------------------------------------------------------------------- |
| **Mode**         | Automation, weekly on a cron trigger. A one-off session also works.          |
| **Project**      | Required — one project per run. Trace access is provisioned for you.         |
| **GitHub skill** | Optional. Add one to let the agent open a PR for a routing or config change. |

You can add context about your cost constraints and approved models; it refines the analysis without overriding the rules above.

Most runs reconfirm existing issues and file nothing new, so a run that reports nothing is a valid result rather than a failure.

## Related docs

* [Signal](/docs/ax/observe/signal) — Ranked correctness and reliability issues
* [Build your own agent](/docs/ax/agents/build-your-own-agent) — The gallery, presets, and customization
* [Connect your harness](/docs/ax/agents/connect-your-harness) — Credentials, skills, and permissions
* [Track costs](/docs/ax/instrument/track-costs) — Get token and cost attributes onto your spans