VS Code

VS Code has no built-in setting for a custom model endpoint — the AI features come from extensions, and each one has its own provider configuration. All of the extensions below accept an OpenAI-compatible endpoint, which is exactly what LuluTokens serves.

Whichever extension you use, the two values are always the same:

Base URL — https://api.lulutokens.ai/v1
API key — your sk-... key from the console

Cline

Open the provider settings

Click the Cline icon in the activity bar, then the gear icon → API Configuration.

Choose "OpenAI Compatible"

Field	Value
API Provider	`OpenAI Compatible`
Base URL	`https://api.lulutokens.ai/v1`
API Key	your `sk-...` key
Model ID	any ID from `/v1/models`, e.g. `deepseek-v4-pro`

Turn on the capabilities your model has

Cline exposes checkboxes for image input and computer use. Leave them off unless the model you picked actually supports them — enabling them otherwise produces requests the provider rejects.

Roo Code

Roo Code is a Cline fork and is configured the same way: Settings → Providers → API Provider: OpenAI Compatible, then fill in the same three values. Roo Code keeps a separate configuration per mode, so set it once and select that profile for Code, Architect and Ask.

Continue

Continue is configured from a file rather than a settings pane.

~/.continue/config.yaml

models:
  - name: DeepSeek V4 Pro (LuluTokens)
    provider: openai
    model: deepseek-v4-pro
    apiBase: https://api.lulutokens.ai/v1
    apiKey: ${{ secrets.LULUTOKENS_API_KEY }}
    roles:
      - chat
      - edit
      - apply

Add one entry per model you want in the picker. provider: openai is correct here — it selects the OpenAI wire format, and apiBase is what sends the request to us instead of to OpenAI.

GitHub Copilot

Copilot Chat can be pointed at a custom OpenAI-compatible endpoint in recent versions ("bring your own key"), but the setting has moved between releases and does not cover inline completions.

Troubleshooting

Symptom	Fix
Extension lists no models	Type the model ID by hand — most extensions do not call `/v1/models`
`404` on every request	Base URL is missing `/v1`, or has a trailing slash the extension does not strip
`400` mentioning images	The model does not accept image input; turn off the vision toggle
Works in chat, fails on edits	The model was not given the `edit`/`apply` role (Continue)

On this page