NEW

Jev is now available

Model prices
View prices Contact sales Telegram support
Help
Support Channel Feedback Telegram
EN
EN English RU Русский 中文 中文 한국어 한국어 TR Türkçe
Log in Try

Connect Cline, Roo Code or Kilo Code to ApiClo

For Cline, Roo Code and Kilo Code, select the OpenAI-compatible provider, enter https://apiclo.com/v1, an ApiClo API key and an exact model ID. The extension sends model requests to ApiClo; any separate extension subscription remains a separate product.

Before you start

  • Install the chosen VS Code extension and create a separate ApiClo key.
  • Look up the model ID and input, output, cache-read and cache-write rates in ApiClo's model table.
Create an API key

Steps

  1. Select a provider

    Open the extension's provider settings and choose OpenAI Compatible. Labels vary by extension and version; choose the option accepting your own Base URL.

  2. Enter the connection

    Use https://apiclo.com/v1, your ApiClo key and a model ID such as claude-sonnet-5. Do not add /chat/completions to the Base URL or an apiclaude/ prefix to the model ID.

  3. Test the workflow

    Send a short message, then test one file-reading tool in a disposable project. Confirm both requests in ApiClo Logs before a long agent task.

Examples

OpenAI Compatible

Provider: OpenAI Compatible
Base URL: https://apiclo.com/v1
API key: YOUR_APICLO_KEY
Model ID: claude-sonnet-5

Check the result

The extension returns a reply, the requested model appears in Logs, and the tool test completes. Check billed tokens and actual cache hits there; entering cache prices in a client does not enable caching.

Troubleshooting

  • 401: re-enter the complete key without spaces and confirm it is active.
  • Model not found: use the ID from ApiClo's table, not its display name or another provider's prefix.
  • Text works but tools fail: inspect the extension version, model capabilities and tool error. Do not assume OpenAI compatibility guarantees every agent feature.

Compatibility and billing

ApiClo bills the model requests it receives. Client-side estimates can differ from billed usage; consult Logs and current rates. Context size, images, tool calls and caching depend on the chosen model and client.

All documentation

Sources