Connect Cursor, Cline, and Continue.dev to Any Model with One Key

The problem: one editor, one provider, one USD bill

Cursor, Cline, and Continue.dev all support "bring your own API key" configurations, which is exactly the flexibility Indian developers need — except the default setup still means signing up for OpenAI or Anthropic billing directly, in USD, with an international card.

Since all three editors support custom OpenAI-compatible endpoints, you can point them at AICredits instead: one key, INR billing via UPI, and access to every major model without juggling separate provider accounts.

Cursor

Open Cursor Settings → Models, scroll to API Keys, and toggle Override OpenAI Base URL. Set:

Base URL: https://api.aicredits.in/v1
API Key:  sk-your-aicredits-key

Add a custom model name if the one you want isn't in Cursor's default list — for example anthropic/claude-3-5-sonnet-20241022 or deepseek/deepseek-chat. Cursor sends requests in the standard OpenAI chat format, which AICredits accepts directly.

Cline (VS Code extension)

In Cline's settings panel, set the API provider to OpenAI Compatible, then fill in:

Base URL: https://api.aicredits.in/v1
API Key:  sk-your-aicredits-key
Model ID: openai/gpt-4o  (or any supported model ID)

Cline reads the model ID directly from this field, so you can switch models by editing one line — no need to reconfigure the provider.

Continue.dev

Continue.dev uses a config.yaml (or the legacy config.json) in ~/.continue/:

models:
  - name: AICredits GPT-4o
    provider: openai
    model: openai/gpt-4o
    apiBase: https://api.aicredits.in/v1
    apiKey: sk-your-aicredits-key
  - name: AICredits Claude Sonnet
    provider: openai
    model: anthropic/claude-3-5-sonnet-20241022
    apiBase: https://api.aicredits.in/v1
    apiKey: sk-your-aicredits-key

Define multiple entries with different model values under the same apiKey and apiBase — Continue.dev's model switcher then lets you flip between them mid-session, all billed from the same INR wallet.

Managing cost while vibe coding

Coding assistants can generate a lot of token volume fast, especially with large file contexts sent on every completion. Two habits keep this predictable:

• Set a per-key budget cap in the AICredits dashboard so an editor stuck in a retry loop can't run past a fixed ₹ limit.
• Use a cheaper model for autocomplete-style suggestions (Gemini 2.0 Flash or GPT-4o-mini) and reserve Claude Sonnet or GPT-4o for larger refactors and multi-file reasoning.

Frequently Asked Questions

Do I lose any editor features by switching the base URL?

No — Cursor, Cline, and Continue.dev all talk to the OpenAI chat completions format, which AICredits implements directly. Autocomplete, chat, and agent modes work the same way.

Can I use different models for different tasks in the same editor?

Yes. In Continue.dev this is built in via named model configs. In Cursor and Cline, switch the model field in settings when you want a different one — the same API key and wallet cover all models.

Does this work with Windsurf or other AI editors too?

Yes, if the editor supports a custom OpenAI-compatible base URL — which most modern AI coding tools do. The setup pattern is the same: base URL + API key + model ID.

Get started

• Create a free account and generate an API key for your editor.
• Read the quickstart guide for supported model IDs.
• Also see Vibe Coding in India: Get LLM API Access Without an International Card.