Cline and Continue custom API: use Claude, GPT and DeepSeek in VS Code

Cline and Continue are two popular AI coding extensions for VS Code. Both support OpenAI-compatible API providers, which means you can point them at apillms and use Claude, GPT, DeepSeek, Grok or Kimi models — all through a single API key and base URL. This guide covers every step for both extensions, from account setup to troubleshooting.

Prerequisites

• An apillms account — sign up to get 50 free starter credits, enough to test several models.
• Your API key, visible on the API keys page (one key per account, created automatically at sign-up).
• VS Code with the Cline and/or Continue extension installed from the VS Code Marketplace.
• A model id you want to use. Some good options for coding work:

The full list of available model ids is on the Models & API page. For a side-by-side comparison of model strengths for coding, see Claude vs GPT vs DeepSeek vs Grok for coding.

Test the model in Playground first

Before wiring up an editor extension, confirm that your key and model id both work. This takes about 30 seconds and saves you from chasing config typos later.

1. Log in to your console and open Playground.
2. Pick the model from the dropdown — for example, claude-sonnet-5.
3. Optionally set a system prompt (e.g. You are a senior Python developer.).
4. Set temperature and max output tokens if you like, or leave the defaults.
5. Type a test message: Write a Python function that reverses a string.
6. Hit send. If you get a response, the same model id and key will work in Cline or Continue.

You can also use the Code button in Playground to see the equivalent cURL, Python, or Node request — useful if you want to verify your setup from a terminal first:

curl https://apillms.com/v1/chat/completions \
  -H "Authorization: Bearer YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-5",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

Set up Cline

Cline (also known as Roo Code) is an autonomous coding agent that can read and edit files, run terminal commands, and iterate on tasks. Here is how to connect it to apillms.

1. Open VS Code and click the Cline icon in the sidebar to open the Cline panel.
2. Click the gear icon (settings) at the top of the Cline panel.
3. Set API Provider to OpenAI Compatible.
4. In the Base URL field, enter:

   https://apillms.com/v1

5. In the API Key field, paste the key from your API keys page. Use the Show key button to reveal and copy it.
6. In the Model ID field, type the exact model id — for example claude-sonnet-5, gpt-5.5, or deepseek-v4-flash. No vendor prefix is needed; just the id as shown on the Models page.
7. Save the settings.

Now open the Cline chat and give it a small task to confirm the connection:

Create a file called hello.py that prints "Hello from Cline" and run it.

If Cline responds and executes the task, you are set. The full walkthrough is on the Cline / Roo Code integration guide.

Switching models in Cline

To try a different model, go back to Cline settings and change only the Model ID field. The base URL and API key stay the same. For example, switch from claude-sonnet-5 to deepseek-v4-flash for cheaper exploratory work, or to claude-opus-5-5 for harder reasoning tasks.

Set up Continue

Continue is an open-source AI code assistant for VS Code and JetBrains. It supports autocomplete, inline editing, and chat — all configurable per model.

Edit config.yaml

Continue stores its provider configuration in a config.yaml file. Open it from the Continue sidebar (click the gear icon), or find it at ~/.continue/config.yaml on macOS/Linux or %USERPROFILE%\.continue\config.yaml on Windows.

Add a model block like this:

models:
  - name: Claude Sonnet 5
    provider: openai
    model: claude-sonnet-5
    apiBase: https://apillms.com/v1
    apiKey: YOUR_KEY

Replace YOUR_KEY with the key from your API keys page. You can add multiple models under the models: list:

models:
  - name: Claude Sonnet 5
    provider: openai
    model: claude-sonnet-5
    apiBase: https://apillms.com/v1
    apiKey: YOUR_KEY

  - name: GPT 5.5
    provider: openai
    model: gpt-5.5
    apiBase: https://apillms.com/v1
    apiKey: YOUR_KEY

  - name: DeepSeek v4 Flash
    provider: openai
    model: deepseek-v4-flash
    apiBase: https://apillms.com/v1
    apiKey: YOUR_KEY

Using the older config.json format

If your Continue installation still uses config.json, the equivalent entry looks like this:

{
  "models": [
    {
      "title": "Claude Sonnet 5",
      "provider": "openai",
      "model": "claude-sonnet-5",
      "apiBase": "https://apillms.com/v1",
      "apiKey": "YOUR_KEY"
    }
  ]
}

Test the connection

1. Save the config file.
2. Reload the Continue panel (or reload the VS Code window).
3. Select the new model from the model picker dropdown in the Continue chat.
4. Send a test prompt, for example: Explain what a Python decorator does in two sentences.

If you get a response, the setup is complete. Step-by-step details and the latest config format are on the Continue integration guide.

Cost estimates for agent and chat use

Both Cline and Continue send system and file-context tokens on every turn, similar to other coding agents. Prompt caching is on automatically for apillms — repeated context across turns is billed at the lower cache-read price, not the full input price. No configuration is needed.

Here is what a typical agent turn costs (100,000 cache-read + 5,000 fresh input + 1,000 output tokens):

For lighter chat-style use in Continue (1,000 input + 500 output tokens per message):

Credits never expire and failed requests are not billed. You can monitor your spend from the console's Overview and Usage pages, filtered by day, week, or month. For a live price comparison with OpenRouter, see the comparison page. For more on how caching reduces costs, read Prompt caching explained.

Troubleshooting

• 401 / Unauthorized: re-copy the key from API keys using the Show key button. If you previously rolled the key, the old one is permanently disabled.
• Model not found: check the model id matches one on the Models & API page exactly. Do not add vendor prefixes like anthropic/ — just use claude-sonnet-5.
• Works in Playground but not in the extension: verify the base URL is exactly https://apillms.com/v1 with no trailing slash. Make sure the provider type is set to OpenAI Compatible (Cline) or openai (Continue), not an Anthropic-native provider.
• Continue shows the model but gives empty responses: check that apiBase (not apiUrl or baseUrl) is the field name in your config. Continue specifically expects apiBase.
• Slow or hanging requests: try a lighter model like deepseek-v4-flash to rule out long generation times on large prompts. Also check that your VS Code has network access (firewalls, proxies).
• Want to verify from the terminal: run the cURL example from the Playground section above. If it returns a valid JSON response, the issue is in the extension config, not the API.

Using Cline and Continue alongside other tools

Because both extensions use the same base URL and key, you can run them side by side in VS Code and use each for different tasks — for example, Continue for quick inline completions and Cline for larger autonomous tasks. The same key also works with other apillms integrations:

• VS Code Chat — use the built-in chat panel with one command, no Copilot needed.
• Cursor — same OpenAI-compatible base URL in Cursor's settings.
• Claude Code — terminal-based agent using the Anthropic-compatible base URL (https://apillms.com, no /v1).

You can track which tool is consuming the most tokens from the console's Usage page, and top up credits any time from the Billing page using PayPal or card.

Takeaway

Both Cline and Continue connect to apillms through the OpenAI-compatible base URL https://apillms.com/v1 and your single API key. The setup for each extension takes under a minute: set the provider to OpenAI Compatible, paste the base URL and key, enter a model id, and send a test message. Test in Playground first to rule out account issues. Swap model ids freely — the base URL and key never change.

FAQ

Do Cline and Continue need different API keys?

No. Both use the same single apillms API key from your account. There is no multi-key feature — just Show key or Roll key on the API keys page.

Which base URL do I use for Cline and Continue?

Use the OpenAI-compatible base URL: https://apillms.com/v1. The Anthropic-compatible URL without /v1 is for Claude-native clients like Claude Code or Claude Desktop.

Can I switch models without reconfiguring everything?

Yes. Change only the model id field in Cline's settings or Continue's config.yaml. The base URL and API key stay the same for all models.

Why test in Playground before configuring the extension?

It confirms your key and model id work independently of the extension's config. If something fails afterward, you know it is an extension setting issue, not an account or API problem.