CANT connect and use chat gptplus account

[rare meteor]

To help others find answers, you can mark your question as solved via Right click solution message -> Apps -> ✅ Mark Solution

[hearty linden]

Let me check the docs and the current source so I don’t guess the wrong thing.

That’s frustrating. A valid Codex OAuth token plus a 404 usually means routing/config mismatch, not “you ran out of credits”.

The 2 most common causes are:

1. Using the wrong route

   • ChatGPT/Codex sign-in works with openai-codex/*
   • openai/* needs a normal OpenAI API key

2. A stale legacy override

   • Old models.providers.openai-codex config can shadow the built-in Codex OAuth path

First, please paste:

openclaw --version
openclaw status

openclaw models status
openclaw doctor

Also tell me which model/ref you’re actually using right now:

• openai-codex/...
• or openai/...

Relevant docs:

• https://docs.openclaw.ai/help/faq#why-does-chatgpt-gpt-54-not-unlock-openaigpt-54-in-openclaw
• https://docs.openclaw.ai/providers/openai
• https://docs.openclaw.ai/gateway/doctor

If you paste those outputs, I can usually pinpoint this pretty fast.

[distant obsidian]

i thinks is good now but extremly slow is normal with codex gpt 5.4 ?

[hearty linden]

Let me check the current docs so I don’t hand-wave this.

Yeah, somewhat normal, especially if you’re on openai-codex/... and doing coding/tool-heavy turns.

A few practical tips:

• Try /fast on in this thread. OpenClaw maps that to OpenAI priority processing for both openai/* and openai-codex/*.
• If you want faster general chat, use /model gpt-5.4 for normal tasks, then switch back to /model openai-codex/gpt-5.4 for coding.
• Very long threads / lots of context can also make it feel slower.

Docs:

• https://docs.openclaw.ai/help/faq#how-do-i-configure-fast-mode-for-gpt-54
• https://docs.openclaw.ai/providers/openai

If it’s not just “a bit slow” but like 30 to 90s before first output every time, paste:

openclaw --version

openclaw status --all

and I’ll help check whether it’s normal slowness or something misconfigured.

You’re close, the problem there is just the command path. Let me verify the right version commands so I don’t send you on a goose chase.

Good news, that error is not the Codex problem. It just means the command path is wrong.

You ran:

node ~/Bureau/openclaw/openclaw --version

…but that file does not exist.

Try this instead:

openclaw --version

If you are running from a source checkout in ~/Bureau/openclaw, then use the built entrypoint from the repo root:

cd ~/Bureau/openclaw
node dist/index.js --version

If openclaw --version says command not found, paste these too:

which openclaw
cd ~/Bureau/openclaw && ls dist/index.js package.json

Once version works, run:

openclaw status --all

openclaw models status

and paste the output. Then we can check whether the remaining slowness is just normal Codex behavior or a real config issue.

Docs:

• https://docs.openclaw.ai/install
• https://docs.openclaw.ai/gateway/troubleshooting

[distant obsidian]

It's really strange, I have to reset the OpenClaw page to get a response...

[hearty linden]

That sounds less like “Codex is slow” and more like the dashboard/UI is losing its live connection. I’m checking the docs for the cleanest way to confirm that.

Yeah, that’s a really useful clue.

If you only get the reply after refreshing the page, that usually means the UI lost its WebSocket connection to the gateway, not that Codex itself is failing. In other words, the answer may already exist, but the page stopped receiving live updates.

Please check these first:

openclaw gateway status
openclaw status --all
openclaw logs --follow

Also try:

openclaw dashboard