I Installed Clawdbot on an Ubuntu Server and Used ChatGPT OAuth as the AI Model

Ethan · January 27, 2026, 1:51am

I recently installed Clawdbot on an Ubuntu server and connected it to ChatGPT using OAuth as the AI model. Since I’m already a ChatGPT Plus subscriber, the setup was surprisingly smooth.

On my Ubuntu server (via SSH), I ran:

curl -fsSL https://clawd.bot/install.sh | bash

After installation, Clawdbot shows a list of available AI models:

Because I’m subscribing to ChatGPT Plus, I simply chose ChatGPT OAuth as the model provider—no API key needed:

Honestly, the $20 ChatGPT Plus subscription is totally worth it. I already use CodeX for vibe coding, and now the same ChatGPT models can even be used directly inside Clawdbot.

Once authenticated, ChatGPT becomes available in Clawdbot:

And yes—GPT-5.2 is selectable and working :

For the messaging channel, I think WhatsApp is the most convenient option. You only need to scan a QR code with your phone, and then you can chat with Clawdbot directly from WhatsApp.

Phone-side setup is straightforward:

To use the Clawdbot web UI dashboard, I did the following:

On the Ubuntu server , run: clawdbot gateway --port your_port --verbose
On my local computer, create an SSH tunnel:

ssh -i /path/to/your_private_key \
  -N -L port:127.0.0.1:port \
  ubuntu@server_ip

Open the dashboard in your browser: http://127.0.0.1:port/?token=xxx

Now you can chat with it on your whatsapp:

You can also assign it cron tasks by chatting too

Ethan · January 27, 2026, 9:24am

I asked it to install Chromium and log in to a website using a QR code. It took a screenshot and sent me the image, which I then scanned with my phone to log in

Ethan · January 28, 2026, 5:05am

I used a Google AI Studio API key with the gemini-3-flash model.

The first chat request worked correctly, but the second request failed with a quota error.

The API returned the following message, indicating that I exceeded the free-tier input token limit.

Error Message

{
  "error": {
    "code": 429,
    "message": "You exceeded your current quota, please check your plan and billing details. For more information on this error, head to: https://ai.google.dev/gemini-api/docs/rate-limits. To monitor your current usage, head to: https://ai.dev/rate-limit.\n* Quota exceeded for metric: generativelanguage.googleapis.com/generate_content_free_tier_input_token_count, limit: 250000, model: gemini-3-flash\nPlease retry in 38.989460429s.",
    "status": "RESOURCE_EXHAUSTED",
    "details": [
      {
        "@type": "type.googleapis.com/google.rpc.Help",
        "links": [
          {
            "description": "Learn more about Gemini API quotas",
            "url": "https://ai.google.dev/gemini-api/docs/rate-limits"
          }
        ]
      },
      {
        "@type": "type.googleapis.com/google.rpc.QuotaFailure",
        "violations": [
          {
            "quotaMetric": "generativelanguage.googleapis.com/generate_content_free_tier_input_token_count",
            "quotaId": "GenerateContentInputTokensPerModelPerMinute-FreeTier",
            "quotaDimensions": {
              "location": "global",
              "model": "gemini-3-flash"
            },
            "quotaValue": "250000"
          }
        ]
      },
      {
        "@type": "type.googleapis.com/google.rpc.RetryInfo",
        "retryDelay": "38s"
      }
    ]
  }
}

Ethan · January 28, 2026, 5:42am

I’m trying to configure Copilot models, but it’s not working. After that failure, I can’t switch to Codex from the Telegram slash command, so I have to run this on my server:

clawdbot models set openai-codex/gpt-5.1-codex-mini

Ethan · January 28, 2026, 8:44pm

Armstrong · January 30, 2026, 7:45am

Today I ran into an issue where I couldn’t use the Copilot model. Every request was returning a 400 error.

I tried using /model in Telegram to switch to Codex. When I checked ~/.clawdbot/clawdbot.json, it showed that the default model had already been set to Codex. However, when I sent a simple hi message in Telegram, I still got a 400 error. Even after running /model, the bot kept saying I was still using Copilot.

To fix this, I ran /reset. After that, the bot responded with:

New session started · model: openai-codex/gpt-5.1-codex-mini

And then everything started working normally again.

Also, another useful slash command is /think. You can set it to off, minimal, low, medium, or high. For complex tasks, high should works well. If you want to save tokens, minimal is a good choice. In my experience, off doesn’t really work as expected, so it’s better to just leave it alone.

Armstrong · January 31, 2026, 12:29am

Using GitHub Copilot Models in OpenClaw

I’ve been using GitHub Copilot with OpenClaw recently, and here’s how the Copilot model usage works.

In OpenClaw , models under github-copilot/* follow GitHub Copilot’s official model catalog and premium-request accounting .

How to interpret Copilot model usage

Chat models (included on paid plans) → Models with a premium multiplier = 0 → They do NOT consume your premium request allowance.
Premium models → Models with a premium multiplier > 0 → They DO consume premium requests.

Non-premium “Chat” models (on paid Copilot plans)

According to the official Copilot model list:

The following models are effectively non-premium chat models on paid Copilot plans:

github-copilot/gpt-4.1
github-copilot/gpt-4o
github-copilot/gpt-5-mini

Can you use GPT-5-mini on a free Copilot account?

Yes — but with limits .

GPT-5-mini usage by plan

Plan	GPT-5-mini Chat Usage
Free	Up to 50 messages per month
Paid	Unlimited, no premium cost

So if you’re on the free GitHub Copilot plan, you can use GPT-5-mini, but it’s capped at 50 chat messages per month. On a paid plan, it’s unlimited and doesn’t count against premium usage.