Fast mode is not working in Codex App Automations

What version of the Codex App are you using (From “About Codex” dialog)?

26.601.21317

What subscription do you have?

Pro

What platform is your computer?

Darwin 23.4.0 arm64 arm

What issue are you seeing?

Codex App automations are not using Fast mode anymore.

I’m using the latest Codex App on macOS. When I turn on Fast mode in the Codex App UI, the app updates ~/.codex/config.toml correctly. I can see service_tier change to "priority" when Fast mode is on, and back to "default" when Fast mode is off.

Fast mode works for my normal Codex session after I turn it on. The problem is with Automations. When a new automation runs, it does not seem to use Fast mode. It feels like it is running in the normal/default speed mode instead.

There is no Speed or Fast mode option in the Automations settings, so I expected automations to use the global Fast mode setting from the app. This is how it worked for me before: when I turned on Fast mode in the app, new automation runs also used Fast mode.

After a recent Codex App update, that seems to have changed. Now the regular session uses Fast mode, but new automation runs do not.

What steps can reproduce the bug?

1. Open Codex App on macOS.

1. In the Codex App UI, turn on Fast mode.

1. Open ~/.codex/config.toml and confirm that Fast mode was saved:

service_tier = "priority"

1. Start a normal Codex session using gpt-5.5 with xhigh reasoning.

1. Confirm that the normal session is using Fast mode.

1. Create or use an existing Automation in the Codex App.

1. Trigger the Automation manually, or wait for it to run on its schedule.

1. Check the new Automation run.

1. Notice that the Automation run does not appear to use Fast mode. It behaves like it is running in the normal/default speed mode instead.

1. Go back to ~/.codex/config.toml and confirm that service_tier is still set to "priority".

1. Optional check: turn Fast mode off in the Codex App UI and confirm service_tier changes back to "default", then turn it on again and confirm it changes back to "priority". The normal session follows the setting, but new Automation runs still do not.

What is the expected behavior?

When Fast mode is turned on in the Codex App, new Automation runs should also use Fast mode.

Additional information

Model: gpt-5.5
Reasoning effort: xhigh
Fast mode behavior:

• UI toggle changes ~/.codex/config.toml service_tier between "priority" and "default"
• Main session runs in Fast mode
• Newly triggered automations run in Standard/default instead