[Codex Vs Code Extension] Reasoning effort when using Custom API Key is not working/show up in ui

Resolved 💬 7 comments Opened Aug 28, 2025 by beforeyouleavethisstayandcalm Closed Aug 29, 2025

What version of Codex is running?

codex-cli 0.25.0

Which model were you using?

gpt-5

What platform is your computer?

Microsoft Windows NT 10.0.26100.0 x64

What steps can reproduce the bug?

  1. Run Codex Extension
  2. in config.toml i set model_reasoning_effort = "high"
  3. the model thinks just as quickly over a complex task, when the reasoning effort is high it thinks for 2-3 minutes

What is the expected behavior?

_No response_

What do you see instead?

_No response_

Additional information

_No response_

View original on GitHub ↗

7 Comments

Remedy92 · 10 months ago

@Kaneki-hash could you actually describe the bug?

beforeyouleavethisstayandcalm · 10 months ago

@Remedy92 I'm using Codex Extension with my own OpenAI API key. I set model_reasoning_effort = "high" in the config.toml file, expecting the model to take 2-3 minutes thinking through complex tasks. But it doesn't work - the model still responds instantly like the reasoning effort is set to low.

It looks like when you use a custom OpenAI API key, the reasoning_effort parameter from config.toml is completely ignored. The model should be thinking longer with high reasoning effort, but instead it just gives quick responses.

daniel-orr · 10 months ago

@Kaneki-hash can you provide the steps to input an OpenAI API key with the extension? I'll attempt to reproduce

beforeyouleavethisstayandcalm · 10 months ago

It is fixed now with 0.4.0 version. Thanks.

daniel-orr · 10 months ago

@Kaneki-hash would you mind sharing your chatgpt.config (obviously remove your key)? I'm attempting to get my key working and not sure if I'm doing it incorrectly. Codex extension won't initialize after I updated my config.

beforeyouleavethisstayandcalm · 10 months ago
@Kaneki-hash would you mind sharing your chatgpt.config (obviously remove your key)? I'm attempting to get my key working and not sure if I'm doing it incorrectly. Codex extension won't initialize after I updated my config.

I didn't have a key initialization stage; the env - OPENAI_API_KEY variable is responsible for that. That is, during initialization, you will be told if there is no OpenAI key to specify it in the variables.

perfumescent · 10 months ago

for v0.4.0, I have set up the config.toml but it still required login with chatgpt.

Finally worked out by setting env var for VS Code, see

https://community.openai.com/t/vscode-extension-with-api-key/1355263/5?u=milkkk_y