🐫 CaMeL Hub
中文 Console

Cline + CaMeL Hub: Connect GPT, Claude & Gemini in VS Code

Cline is an open-source autonomous coding agent that runs as a VS Code extension (also installable in Cursor and other VS Code-compatible editors). It reads and edits files across your project, runs terminal commands, and works through multi-step tasks on its own — asking for your approval before each file write or command unless you turn on auto-approve. Cline doesn't lock you into one model provider: under the hood it just calls a chat-completions API, so pointing it at CaMeL Hub through the built-in "OpenAI Compatible" provider gives you one API key for GPT-5.5, Claude Sonnet 5, Gemini 3.1 Pro, DeepSeek V4 and 480+ other models, and you can switch models per task without touching credentials again.

Cline ↗

  1. Install the Cline extension

    In VS Code, open the Extensions view (Ctrl/Cmd+Shift+X), search for "Cline", and install it. Cline adds an icon to the activity bar that opens its chat panel.

  2. Open Cline's API provider settings

    Click the gear/settings icon inside the Cline panel (or run "Cline: Open Settings" from the Command Palette). Under API Configuration, set the provider dropdown to "OpenAI Compatible" — this is the generic mode Cline uses for any endpoint that speaks the OpenAI chat-completions format, including CaMeL Hub.

  3. Fill in Base URL, API Key and Model ID

    Enter the CaMeL Hub API base (full path, including the /v1 suffix — Cline's OpenAI-compatible client appends /chat/completions to whatever you enter, so the bare host alone will 404), paste an API key generated from the console's Keys page, and type the model ID exactly as it appears on CaMeL Hub.
    Base URL:  https://api.camel-hub.com/v1
API Key:   <your CaMeL Hub API key>
Model ID:  claude-sonnet-5

  4. Generate an API key

    In a browser, go to the CaMeL Hub console's token page, create a new key, and copy it — it's shown only once. Paste it into the API Key field from the previous step.

  5. (Optional) Verify the key with a raw request

    Before trusting it inside Cline, confirm the key and model work with a plain curl call. A successful response returns a normal chat-completion JSON object; a 401 means the key is wrong or was pasted with extra whitespace, a 404 usually means the Base URL is missing /v1.
    curl https://api.camel-hub.com/v1/chat/completions \
  -H "Authorization: Bearer $CAMEL_HUB_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-5",
    "messages": [{"role": "user", "content": "ping"}]
  }'

  6. Save and start a task

    Save the settings, open a project folder, and give Cline a task in plain language (e.g. "add input validation to the signup form and write a test for it"). Cline will propose a plan, then read/edit files and run commands step by step, pausing for your approval at each change unless auto-approve is enabled.

  7. Switch models per task

    Because the field is a free-text Model ID rather than a fixed dropdown, you can swap it any time: a fast, inexpensive model like deepseek-v4-flash for small mechanical edits, and a stronger reasoning model like claude-sonnet-5 or gpt-5.5 for architecture-level changes — without creating a second account or key.

Frequently asked questions

Cline doesn't show a model list to pick from — is that a problem?

No. The "OpenAI Compatible" provider in Cline is a generic passthrough, so it never auto-fetches a model catalog the way built-in providers do — you always type the model ID by hand. Copy the exact model name from the CaMeL Hub models page (for example claude-sonnet-5, gpt-5.5, gemini-3.1-pro, or deepseek-v4-flash) rather than guessing at spelling.

Cline says the API key or request failed — what should I check first?

Three usual causes: the Base URL is missing the /v1 suffix (Cline's OpenAI-compatible client won't add it for you), the API key was copied with a trailing space or newline, or the model ID doesn't match a model available on your account. Testing the same key and model with a plain curl request (see the verification step above) isolates whether the problem is in Cline's settings or the key itself.

Which model should I pick for coding tasks in Cline?

For everyday feature work and debugging, claude-sonnet-5 or gpt-5.5 give the most reliable multi-step tool use. For quick, low-risk edits (renames, small refactors, boilerplate), a cheaper model like deepseek-v4-flash keeps token spend down. There's no wrong choice technically — Cline works the same way regardless of model — it's purely a cost/quality tradeoff you control per task.

Do I need a separate account for each model or provider?

No. One CaMeL Hub API key already authenticates every model on the platform, so switching the Model ID field in Cline is the only step needed to move between GPT, Claude, Gemini and DeepSeek — there's nothing else to reconfigure.