Subagent Pricing Incorrect

Support Bug Reports
,
1

Where does the bug appear (feature/product)?

Cursor IDE

Describe the Bug

From what I can see, even if a sub agent uses a different model to execute, i.e. GPT-5.2 is the main model and sub agent is set to use Grok Code, the usage for GPT5.2 is increased and not Grok Code when executed by GPT-5.2.

Steps to Reproduce

Create a local subagent using a smaller model. Execute it using a larger model.

Expected Behavior

Billed for tokens used in larger model context window at rates of the larger model and billed for tokens used in the smaller model context window at rates of the smaller model

Operating System

Windows 10/11

Version Information

Version: 2.5.0-pre.36.patch.0 (user setup)
VSCode Version: 1.105.1
Commit: 056d4e6cf618e13bbb1a7dda5b6e5ea78dde2dd0
Date: 2026-02-01T08:32:37.526Z
Build Type: Stable
Release Track: Nightly
Electron: 39.3.0
Chromium: 142.0.7444.265
Node.js: 22.21.1
V8: 14.2.231.22-electron.0
OS: Windows_NT x64 10.0.26100

For AI issues: which model did you use?

Models: GPT 5.2 Codex and Grok Code.

For AI issues: add Request ID with privacy disabled

b5990298-94da-4bbc-b26f-23cb5f448597

Additional Information

My only visibility on token usage is on the Billing tab on dashboard, and I do understand tokens to the billing dashboard might be delayed. I tested this in isolation.

Does this stop you from using Cursor

No - Cursor works, but with this issue

4

Hey @Noxide!

Thanks for flagging this. I can confirm that the billing is correct, but the model name is wrong in the billing dashboard. We’re working on a fix for that that should land in 2.5.

6

@Colin I assume the updated stats will then reflect usage correctly?

7

To clarify, all usage data (dollar amounts and requests used) is accurate right now. Only the model name is incorrect.