# Cursor Cost Tracking: What You Can See, What You Can't, and What Cursor Shows

How to track Cursor spend. Cursor's Spending tab, its two usage pools, July's dollar removal, and what ACP adds: tool-call logging always, token pricing only for models you bring your own key for.

<div style="margin:8px 0 28px;padding:20px 22px;border:1px solid var(--line-2);border-radius:12px;background:var(--color-accent-light,#f0effe);">
  <p style="margin:0 0 12px;font-size:15px;line-height:1.6;color:var(--acp-text);"><strong>The limit, first.</strong> ACP's Cursor hook sees tool calls (file edits, shell commands, MCP calls), not tokens. The ACP proxy can only price models you've added to Cursor with your own API key. Cursor's bundled subscription models (Composer, Grok, the included pool) go through Cursor's backend. No base URL you set reaches them, so no outside tool can meter them.</p>
  <p style="margin:0;font-size:15px;line-height:1.6;color:var(--acp-text);">What ACP adds: a tool-call log for every Cursor session, whichever models you use, plus per-call token pricing for any model you've pointed at the proxy with your own key.</p>
</div>

Most cost-tracking pages lead with a number. This one leads with a boundary, because for Cursor the boundary decides everything. What you can meter from outside Cursor depends on whether the model call is yours to route. Cursor's included models never leave Cursor's backend. Models you bring with your own key do, and that is the only door a proxy has.

## What Cursor shows

Cursor's **Spending tab** shows real-time usage for two pools: Cursor Models (Composer, Grok and the rest of what your plan includes) and Other Models (third-party models drawing on the same allowance). It also shows your remaining allowance, any on-demand charges, and the reset date for your billing cycle.

<!-- source: https://cursor.com/help/models-and-usage/usage-limits -->

If you have used Cursor for a while, you will notice a change from July 2026. Self-serve individual and Teams plans stopped showing a dollar figure per request or per model. Rows covered by your plan now say "Included". The per-day dollar graph and the cost column in usage CSVs went with it. Cursor staff confirmed the change applied to old records too: "because this is applied when records are read, historical results are affected, too." The stated reason is that a dollar amount at API rates was often higher than what customers actually paid, which confused people.

Dollar amounts did not disappear everywhere. On-demand spending still shows in dollars, the CSV cost column still carries dollars for on-demand rows, and Enterprise plans kept the full per-model breakdown.

<!-- source: https://www.developersdigest.tech/blog/cursor-removes-dollar-costs-usage-page -->

For a team, the admin dashboard and Admin API go further. They break down spend and tokens by model, token type and developer. That is an Enterprise-tier view, and it is Cursor's own view of Cursor's own traffic. It doesn't tell you what ran, what a hook decided, or what got blocked. That is a different question from what the model calls cost.

## Bring your own key

Cursor lets you add your own API key for OpenAI, Anthropic, Google, Azure or AWS Bedrock models under Settings → Models. Cursor's docs state the limit: **"Custom API keys only work with chat models. Tab completion continues using Cursor's built-in models."** Even with your key, a request is still "routed through Cursor's servers for final prompt building" before it reaches the provider.

<!-- source: https://cursor.com/help/models-and-usage/api-keys -->

This is the model class the ACP proxy can reach. Cursor's model settings include an OpenAI-compatible base URL override that applies to a model you configured with your own key. Point it at `https://api.agenticcontrolplane.com/v1` and authenticate with your ACP workspace key. ACP then prices every call to that model from the usage fields in the response: model, tokens, cached share, dollars.

```bash
curl -sf https://agenticcontrolplane.com/install.sh | bash
# Cursor Settings -> Models -> your BYOK model -> API Base URL:
#   https://api.agenticcontrolplane.com/v1
# Auth: your ACP workspace key (gsk_...)
```

<!-- source: https://agenticcontrolplane.com/integrations/cursor -->

This does not reach Composer, Grok or anything else in Cursor's included pool. Cursor's backend sends those calls as part of your subscription. There is no base-URL field for them because there is no outbound request for it to redirect. If most of your Cursor usage is the bundled models, as it is for most Pro and Team seats, the proxy has nothing to meter.

## A priced BYOK session

<div class="acp-shot-frame">
  <img src="/assets/img/screenshots/session-trace-dark.png" alt="An ACP session trace: each model call priced with its token and cached-token counts, interleaved with the tool calls it drove" style="width:100%;height:auto;border:1px solid var(--line-2);border-radius:10px;box-shadow:0 20px 50px -24px rgba(0,0,0,0.9);margin:8px 0;" />
</div>

This is the trace for a Cursor model you added with your own key and pointed at the ACP proxy. Each model call is priced, next to the tool calls it drove. ACP shows the same trace shape for other harnesses. In Cursor it only exists for models routed through the proxy, never for Composer or Grok on the included pool.

## What the hook sees on every setup

Separately from model metering, [the ACP hook for Cursor](/integrations/cursor) runs on `preToolUse` and `postToolUse`. It logs the tool calls it can see, such as file edits, shell commands and MCP calls, with the decision on each. It does this whichever model produced them, and it needs neither BYOK nor the proxy. Some Cursor built-in tools don't emit these events, so the log can miss them. The [coverage table](/coverage) has the current state.

<!-- source: https://agenticcontrolplane.com/integrations/cursor -->

The hook can also hold a call for approval. In Cursor that is a block: you approve on the console or from the email, then the agent retries. Cursor doesn't show the hold inline the way Claude Code does.

## What you get, by setup

- **Tool-call log, every setup.** File edits, shell, MCP calls, and the decision on each. This is the hook, and it doesn't depend on which models you use.
- **Per-call token pricing, BYOK models only.** Model, input, output and cached tokens, and dollars at list rates, for any model you added with your own key and pointed at the proxy.
- **Nothing per call for the bundled pool.** Composer, Grok and the rest of the included models can't be priced from outside Cursor. Cursor's Spending tab is the source for that traffic: pool totals, remaining allowance and on-demand dollars. Since July it no longer breaks included usage down by request or model.

## Limits

- **The proxy only prices BYOK.** If your Cursor plan doesn't use custom-key models, there is no per-call cost view to add. Say so before anyone sets an expectation the account can't meet.
- **The hook sees tool calls, not tokens**, on every setup.
- **Tab completion is out of reach.** It always uses Cursor's built-in models, so BYOK and the proxy setting never apply to it.
- **Cursor's per-model dollar breakdown is gone for self-serve plans.** If you need that detail on Cursor's own traffic and you aren't on Enterprise, what's left is the token-only Usage page and the on-demand page, which still shows dollars.

## Compared with Claude Code

Claude Code works the other way. Every session, subscription or API key, can be priced because [the ACP hook reads the transcript Claude Code already writes](/blog/claude-code-cost-tracking-reference). No proxy and no BYOK needed. Cursor doesn't hand an outside hook that transcript, so metering depends on whether you can route the model call yourself.

## Frequently asked questions



## More

- [Install ACP for Cursor](/integrations/cursor): the hook, the proxy setting and the documented limits
- [Cursor's full control model](/controls/cursor): Run Modes, the sandbox, and where the model ends
- [Cursor hooks reference](/blog/cursor-hooks-reference): the events, the contract and failClosed
- [Cursor: Bring your own API key](https://cursor.com/help/models-and-usage/api-keys): Cursor's reference for BYOK scope
- [Cursor: Usage and limits](https://cursor.com/help/models-and-usage/usage-limits): the Spending tab, pools and on-demand charges
- [Claude Code cost tracking reference](/blog/claude-code-cost-tracking-reference): a harness where every session is priced
- [Codex CLI cost tracking](/blog/codex-cli-cost-tracking): the sibling reference for Codex
