Codex just not working at all.

Resolved 💬 16 comments Opened Aug 7, 2025 by JarrettOneSource Closed Oct 17, 2025

💡 Likely answer: A maintainer (seratch, member) responded on this thread — see the highlighted reply below.

What version of Codex is running?

codex-cli 0.16.0

Which model were you using?

Any

What platform is your computer?

Windows Subsystem Linux Ubuntu 24.04 LTS

What steps can reproduce the bug?

I install ubuntu distro fresh
I update the OS with build tools and what not
I install nvm and npm and node
I install codex

I run codex and when I say something to the agent nothing happens and after some time it throws the error.

What is the expected behavior?

I expect it to respond saying something.

What do you see instead?

jjohnson@Workstation20:/mnt/c/Users/David Johnson/Documents/GitHub$ codex -m "gpt-5"

>_ You are using OpenAI Codex in /mnt/c/Users/David Johnson/Documents/GitHub

 To get started, describe a task or try one of these commands:

 /init - create an AGENTS.md file with instructions for Codex
 /status - show current session configuration and token usage
 /diff - show git diff (including untracked files)
 /prompts - show example prompts

/status
📂 Workspace
  • Path: /mnt/c/Users/David Johnson/Documents/GitHub
  • Approval Mode: on-request
  • Sandbox: workspace-write

🧠 Model
  • Name: gpt-5
  • Provider: OpenAI
  • Reasoning Effort: Medium
  • Reasoning Summaries: Auto

📊 Token Usage
  • Input: 0
  • Output: 0
  • Total: 0

user
test

🖐 error sending request for url (https://api.openai.com/v1/responses)

Additional information

If I install codex version v0.1.2505172129 it works like a charm on the same machine with nothing else different.

If I install v0.16.0 on Windows directly it works but the commands pop out a new powershell window every single time and it's goofy and the commands don't return anything to the model so it technically fails.

View original on GitHub ↗

16 Comments

4r7 · 11 months ago

I am seeing the same issue in Windows where it will try to run commands in a separate window, then say it can't access a sandbox even though WSL2 is running - have tried both docker and ubuntu backends - tested in their respective terminals and all files can be accessed, but codex always fails.

25May2025 · 11 months ago

I'm also experiencing basic errors. It can write files, but not read them. Just in regular vscode's terminal, not in a dev container or remote server or something. Just my regular desktop.

pdemro · 11 months ago

Just upgraded Codex and not seeing responses whatsoever. The very same endpoint works with Cline. If I ask it about the project, I can see some output of attempting to run file system commands, but that is about it.

0.19.0

JarrettOneSource · 11 months ago

The workaround I found is to uninstall codex and reinstall this specific version until they fix the newer versions.

npm uninstall -g @openai/codex
npm install -g @openai/codex@0.1.2505172129

stickyfingies · 11 months ago

The codex app is unstable across all platforms. I've seen the configuration instructions on the README change several times, which is anecdotal evidence, sure, but you'd think they'd have this software more polished for the gpt-oss release. Of course local LLM users are going to immediately try running OpenAI's OSS model on OpenAI's OSS agentic CLI toolkit. It's a shame the cli toolkit wasn't in a more stable place when they released the model.

pdemro · 11 months ago

Seeing this on Arch Linux as well. This might be a duplicate with https://github.com/openai/codex/issues/1995

pdemro · 11 months ago

Looks like this is solved with 21 for me.

JarrettOneSource · 11 months ago

The issue of it not doing anything is still present with me.

seratch member · 11 months ago

The latest version is 0.22.0 and the version could behave much better. Could you try the latest version and let us know how it goes? You can upgrade the version by npm -g i @openai/codex@latest.

JarrettOneSource · 11 months ago

It is still not working. I updated and sent a simply "Hi!" message and it times out.

jjohnson@Workstation20:/mnt/c/Users/David Johnson/Documents/GitHub/SmartSync2.0$ npm -g i @openai/codex@latest

added 6 packages, removed 217 packages, and changed 5 packages in 7s
jjohnson@Workstation20:/mnt/c/Users/David Johnson/Documents/GitHub/SmartSync2.0$ codex

>_ You are using OpenAI Codex in /mnt/c/Users/David Johnson/Documents/GitHub/SmartSync2.0

 To get started, describe a task or try one of these commands:

 /init - create an AGENTS.md file with instructions for Codex
 /status - show current session configuration and token usage
 /diff - show git diff (including untracked files)
 /prompts - show example prompts

user
Hi!

🖐 error sending request for url (https://api.openai.com/v1/responses)

/status
📂 Workspace
  • Path: /mnt/c/Users/David Johnson/Documents/GitHub/SmartSync2.0
  • Approval Mode: on-request
  • Sandbox: workspace-write
  • Session ID: 8649a652-0672-46d1-8bfa-8ba142662040

🧠 Model
  • Name: gpt-5
  • Provider: OpenAI
  • Reasoning Effort: Medium
  • Reasoning Summaries: Auto

📊 Token Usage
  • Input: 0
  • Output: 0
  • Total: 0

▌ Ask Codex to do anything
 ⏎ send   Ctrl+J newline   Ctrl+C quit                                                                                                                                                                                                                                                                                                                                        ```

seratch member · 11 months ago

thanks for sharing this. just in case, could you share the output of codex --version?

JarrettOneSource · 11 months ago

jjohnson@Workstation20:/mnt/c/Users/David Johnson/Documents/GitHub/SmartSync2.0$ codex --version
codex-cli 0.22.0

JarrettOneSource · 10 months ago

I think the last update that fixed the issue where commands would pop out a separate powershell window has fixed it so I can now run the most recent version directly on my Windows machine.

However the WSL version is still not working for me.

user
Hello
⚠  stream error: error sending request for url (https://api.openai.com/v1/responses); retrying 1/5 in 216ms…

⚠  stream error: error sending request for url (https://api.openai.com/v1/responses); retrying 2/5 in 420ms…

⚠  stream error: error sending request for url (https://api.openai.com/v1/responses); retrying 3/5 in 864ms…

⚠  stream error: error sending request for url (https://api.openai.com/v1/responses); retrying 4/5 in 1.507s…

⚠  stream error: error sending request for url (https://api.openai.com/v1/responses); retrying 5/5 in 3.438s…


🖐  error sending request for url (https://api.openai.com/v1/responses)

▌ Improve documentation in @filename
 ⏎ send   Ctrl+J newline   Ctrl+T transcript   Ctrl+C quit

JarrettOneSource · 10 months ago

Works great on Windows now although the pwsh.exe commands seem to bypass the nice Linux parsing and what not "Exploring ...", "Reading ..." etc.

It still does not work at all on WSL. The older version I referenced earlier did work on WSL.

JarrettOneSource · 9 months ago

Newest version tells me to run it on WSL windows for best performance but it still doesn't work lol

╭──────────────────────────────────────────────────────────╮

  /init - create an AGENTS.md file with instructions for Codex
  /status - show current session configuration
  /approvals - choose what Codex can do without approval
  /model - choose what model and reasoning effort to use
  /review - review any changes and find issues

• Model changed to gpt-5 with high reasoning


› Look at the agent threads html page and ensure that attached images / files are supported on this page. I should be
  able to click to download images and other files from the agent thread page.


⚠️ stream error: error sending request for url (https://api.openai.com/v1/responses); retrying 1/5 in 204ms…

/status

╭──────────────────────────────────────────────────────────────────────────────╮
│  >_ OpenAI Codex (v0.45.0)                                                   │
│                                                                              │
│  Model:            gpt-5 (reasoning high, summaries auto)                    │
│  Directory:        /mnt/c/Users/David Johnson/Documents/GitHub/VanguardAI-v2 │
│  Approval:         never                                                     │
│  Sandbox:          danger-full-access                                        │
│  Agents.md:        AGENTS.md                                                 │
│  Account:          API key configured (run codex login to use ChatGPT)       │
│  Session:          0199bf8b-b5d0-7f43-a30d-83eb8c4ec4ec                      │
│                                                                              │
│  Token usage:      0 total  (0 input + 0 output)                             │
│  Context window:   100% left (0 used / 272K)                                 │
│  Limits:           send a message to load usage data                         │
╰──────────────────────────────────────────────────────────────────────────────╯

◦ Working (52s • esc to interrupt)


› Explain this codebase

  ? for shortcuts

tibo-openai collaborator · 9 months ago

Please open new issues specifically for frictions on windows. Going to mark this one as closed as other issues referenced here should all be resolved.