›_ Codex Issues
Codex Issues / #20162

Speed setting resets to Fast on VS Code reopen and cannot be changed while Codex settings tab is open

Open 💬 20 comments Opened Apr 29, 2026 by shurkanTwo
💡 Likely answer: A maintainer (github-actions[bot], contributor) responded on this thread — see the highlighted reply below.

What version of the IDE extension are you using?

26.5422.71525

What subscription do you have?

Business

Which IDE are you using?

VS Code Version: 1.117.0 (user setup) Commit: 10c8e557c8b9f9ed0a87f61f1c9a44bde731c409 Date: 2026-04-21T16:12:14-07:00 Electron: 39.8.7 ElectronBuildId: 13841579 Chromium: 142.0.7444.265 Node.js: 22.22.1 V8: 14.2.231.22-electron.0 OS:

What platform is your computer?

Windows_NT x64 10.0.26200

What issue are you seeing?

The Codex VS Code extension does not persist my selected speed setting. I set the model to ChatGPT 5.5 with reasoning effort set to Low, and set the speed to Normal. Every time I close and reopen VS Code, the speed setting automatically resets back to Fast.

There is also a second related issue: if the Codex settings tab is already open in VS Code, I cannot even temporarily change the speed setting. When I switch it from Fast to Normal, it immediately changes back to Fast.

What steps can reproduce the bug?

  1. Open VS Code with the Codex extension installed.
  2. Open Codex settings.
  3. Configure: Model: ChatGPT 5.5; Reasoning effort: Low; Speed: Normal
  4. Close and reopen VS Code.
  5. Open Codex settings again.

What is the expected behavior?

The speed setting should remain set to Normal after VS Code is reopened.

Also, while the Codex settings tab is open, changing the speed setting should apply immediately and remain selected unless I change it again.

Additional information

The speed setting resets to Fast every time VS Code is reopened.

Additionally, when the Codex settings tab is open, changing the speed setting from Fast to Normal immediately reverts back to Fast, so the setting cannot be changed even temporarily.

View original on GitHub ↗

20 Comments

github-actions[bot] contributor · 2 months ago

Potential duplicates detected. Please review them and close your issue if it is a duplicate.

  • #20061

Powered by Codex Action

shurkanTwo · 2 months ago

It's not a duplicate. I see the option. However it just resets to Fast when I reopen VS Code.

<img width="1029" height="588" alt="Image" src="https://github.com/user-attachments/assets/9d9fdfbd-604c-4cc4-b69f-c0c847a56ce3" />

egit-m · 2 months ago

I second this, I have the same issue with the VS code extension both in windows and wsl.

acshedley · 2 months ago

Same here, also using VS Code on WSL.

Additionally, disabling Fast Mode in config.toml only appears to affect the Codex CLI. The VSCode Codex extension still defaults to Fast Mode being enabled on every relaunch, despite fast_mode = false being set under the [features] section.

neodon · 2 months ago

Similar issue that just started a couple days ago, except for the Codex App.

It refuses to keep fast mode on between sessions and sometimes within the same session. Also ignores fast_mode = true in config.toml. Probably the same underlying bug introduced with a recent update.

Vacilando · 2 months ago

Same problem in VSCode Codex plugin on Win11 Pro, using model "gpt-5.4".

The Fast mode is ON and there is no way to change it permanently to Standard via settings, and it also ignores fast_mode = false in config.toml. In all cases, after restart of VS Code, it is back on Fast, gobbling up the API credits :-(

ilya-dmitriev-clearscale · 2 months ago

The same issue. I use a business account and try to set up in config.toml option, but it does not work well.

fast_mode = false

shurkanTwo · 2 months ago

I would really like to see this fixed. This makes the Codex extension increasingly unpleasant to use.

mfalkvidd · 2 months ago

I was able to disable fast mode by adding

[features]
fast_mode = false

in ~/.codex/config.toml

This locks the extension to "Standard" speed though. Changing the speed to "Fast" in Codex Settings in VScode will not persist.
Maybe this is the reverse of the problem described in https://github.com/openai/codex/issues/20769

Documentation reference: https://developers.openai.com/codex/config-basic#supported-features

ilya-dmitriev-clearscale · 2 months ago
I was able to disable fast mode by adding `` [features] fast_mode = false `` in ~/.codex/config.toml This locks the extension to "Standard" speed though. Changing the speed to "Fast" in Codex Settings in VScode will not persist. Maybe this is the reverse of the problem described in https://github.com/openai/codex/issues/20769 Documentation reference: https://developers.openai.com/codex/config-basic#supported-features

I tried to set this mode but no luck

Vacilando · 2 months ago

Yes, it seems that the bug has been fixed.

The config / fast_mode = false pin the speed to "Standard", finally.

artemiuss · 2 months ago

not fixed for me

acshedley · 2 months ago

If you're still having issues, try the pre-release version of the extension. It appears to be working correctly

artemiuss · 2 months ago

switching to pre-release version didn't help

WilliamK7 · 2 months ago

I can reproduce this as well in WSL.

I set Agent Speed to Standard in the VS Code extension settings, but after closing/reopening the settings page it reverts to Fast. The /fast slash command also does not persist the change.

The pre-release version extension doesn't solve this problem either.

artemiuss · 2 months ago

You’ve described my exact situation. However, in my case, this issue only occurs on the environment with the corporate subscription; on another computer with my personal ChatGPT Plus subscription, I don’t have this problem. There may be other differences between these environments, but I still haven’t been able to figure out what the problem is. I’ve reinstalled both VS Code and the Codex extension, and I’ve logged in again—all to no avail. It’s frustrating.

WilliamK7 · 2 months ago

They made Codex's Fast service tier the default for eligible ChatGPT plans (enterprise and business-like) according to this issue https://github.com/openai/codex/issues/19230

So maybe there are some underlying bugs.

WilliamK7 · 2 months ago

I found out a potential way to deal with this problem in WSL.

  1. Launch VS Code in pure Windows environment
  2. Open Codex settings
  3. Set the default speed option to be standard instead of fast
  4. The setting is saved and also works in WSL
artemiuss · 2 months ago

Thanks, that helped!

claell · 1 month ago

Adding a Codex Desktop data point after updating on 2026-06-04.

Environment:

  • Codex Desktop Microsoft Store package: OpenAI.Codex_26.601.2237.0_x64__2p2nqsd0c76g0
  • Package version: 26.601.2237.0
  • Platform: Windows, Microsoft Store installation
  • Model/intelligence setting observed: GPT-5.5 High

Observed behavior:

  • After updating Codex Desktop today, the Speed setting was silently set to Fast for GPT-5.5 High.
  • I did not intentionally enable Fast.
  • This is unexpected because Speed directly affects usage/quota behavior, so silently changing it can materially change how much usage a normal workflow consumes.

Expected behavior:

  • Codex Desktop should preserve the existing Speed preference across app updates.
  • If an app update, account eligibility rule, or defaulting/migration path changes Speed to Fast, the app should clearly prompt or notify before applying an increased-usage setting.
  • After manually setting Speed back to Standard, the app should not silently revert it to Fast.