Automation may be using Codex 5.3 high instead of the selected Haiku 4.5

Where does the bug appear (feature/product)?

Background Agent (GitHub, Slack, Web, Linear)

Describe the Bug

Hi Cursor team,

I hope you are doing well.

I am posting to ask about what may be an issue with Automation model selection.

I configured the following Automation to use Haiku 4.5, but from what I can see, it seems to be running with Codex 5.3 high instead.

Details:

  • Automation name: 目標達成OS接続確認
  • Selected model: Haiku 4.5
  • Actual model observed: Codex 5.3 high
  • Trigger type: cron
  • Cron expression: 0 7-23 * * *

Expected behavior:

  • The Automation runs with Haiku 4.5, which is the model I selected.

Observed behavior:

  • The Automation seems to run with Codex 5.3 high instead.

Environment:

  • Cursor Version: 2.6.19 (Universal)
  • VSCode Version: 1.105.1
  • Commit: 224838f96445be37e3db643a163a817c15b36060
  • Date: 2026-03-12T04:07:27.435Z
  • Build Type: Stable
  • Release Track: Default
  • Electron: 39.4.0
  • Chromium: 142.0.7444.265
  • Node.js: 22.22.0
  • V8: 14.2.231.22-electron.0
  • OS: Darwin arm64 25.3.0

Steps:

  1. Create an Automation
  2. Select Haiku 4.5 as the model
  3. Set the cron schedule to 0 7-23 * * *
  4. Wait for the scheduled run
  5. Check the model used in the run result

I was not able to locate a Request ID for this Automation run. If there is a place where I can find it, I would be grateful if you could let me know.

Could you please advise whether this is a known issue, and whether there is any workaround available?

If needed, I can also share screenshots.

Thank you for your time and support.

Steps to Reproduce

Create an Automation

Set the model to Haiku 4.5

Configure the cron schedule as 0 7-23 * * *

Let the Automation run on schedule

Observe that the run appears to use Codex 5.3 high instead of Haiku 4.5

Operating System

MacOS

Version Information

  • Cursor Version: 2.6.19 (Universal)
  • VSCode Version: 1.105.1
  • Commit: 224838f96445be37e3db643a163a817c15b36060
  • Date: 2026-03-12T04:07:27.435Z
  • Build Type: Stable
  • Release Track: Default
  • Electron: 39.4.0
  • Chromium: 142.0.7444.265
  • Node.js: 22.22.0
  • V8: 14.2.231.22-electron.0
  • OS: Darwin arm64 25.3.0

macOS 26.3.1

For AI issues: which model did you use?

Haiku 4.5

Does this stop you from using Cursor

No - Cursor works, but with this issue

Hey, thanks for the detailed report.

We checked the backend data for your automation, and good news, your cron runs are actually executing with Haiku 4.5 as configured. The requested_model for all your scheduled runs shows claude-4.5-haiku-thinking, which is correct.

What you’re likely seeing is a known UI display bug where the automation page shows the default model name (Codex 5.3 high) instead of the model that’s actually being used. The team is aware of this issue.

To help us pin down exactly where the display is wrong, could you share a screenshot showing where you see “Codex 5.3 high”? Specifically:

  • Is it in the automation edit view?
  • Or in the run results or history page?

Your automation is working correctly under the hood, it’s just the label that’s misleading. Let me know if you have any other questions.

This topic was automatically closed 22 days after the last reply. New replies are no longer allowed.