> ## Documentation Index
> Fetch the complete documentation index at: https://docs.aiohub.org/llms.txt
> Use this file to discover all available pages before exploring further.

# Choose a client

> Pick the right AIOHub setup guide for your client or tool.

Choose the page by client protocol. Do not mix protocol-specific base URLs: OpenAI-compatible apps usually use `/v1`, Claude Code and Gemini CLI use the gateway root, and Codex CLI uses `/v1` with Responses.

## Protocol quick reference

| Client or use case        | Protocol                                    | Base URL                       | Common group                                              |
| ------------------------- | ------------------------------------------- | ------------------------------ | --------------------------------------------------------- |
| Claude Code               | Claude Messages                             | `https://api.aiohub.org`       | `anthropic`, `claude-bedrock`, `claude-kiro`              |
| Codex CLI                 | OpenAI Responses                            | `https://api.aiohub.org/v1`    | `codex`, `codex-passthrough`                              |
| OpenCode                  | OpenAI Chat Completions or OpenAI Responses | `https://api.aiohub.org/v1`    | `openai`, `codex`, `opensource`                           |
| Gemini CLI                | Gemini API                                  | `https://api.aiohub.org`       | `gemini`                                                  |
| VS Code (Claude Code)     | Claude Messages                             | `https://api.aiohub.org`       | `anthropic`, `claude-bedrock`, `claude-kiro`              |
| VS Code (Codex)           | OpenAI Responses                            | `https://api.aiohub.org/v1`    | `codex`, `codex-passthrough`                              |
| OpenAI SDK / general apps | OpenAI Chat Completions                     | `https://api.aiohub.org/v1`    | `openai`                                                  |
| Imagine                   | OpenAI Images                               | `https://api.aiohub.org`       | `openai`, `codex*`                                        |
| Claude Cowork 3P          | Claude Messages                             | `https://api.aiohub.org`       | `anthropic`, `claude-bedrock`, `claude-kiro`              |
| Chatbox                   | OpenAI Chat Completions                     | `https://api.aiohub.org/v1`    | `openai`, `auto`                                          |
| Open-source model clients | OpenAI Chat Completions or Claude Messages  | Depends on the client protocol | `opensource`, `opensource-discount`, `opensource-economy` |

<Note>All API keys use the `sk-` format, but assigned groups control visible models, endpoints, and pricing. `auto` is a routing control value, not a fixed group; verify the client, model, and endpoint you plan to use.</Note>

<Columns cols={2}>
  <Card title="Claude Code" icon="https://mintcdn.com/qyhfrankgmailcom/--B4GB7WTjd2TEz0/images/common/clients/logos/claude.svg?fit=max&auto=format&n=--B4GB7WTjd2TEz0&q=85&s=479726235c925d415c203874a908ea34" href="/en/clients/claude-code" width="24" height="24" data-path="images/common/clients/logos/claude.svg">
    Uses Claude Messages. Set `ANTHROPIC_BASE_URL` to `https://api.aiohub.org`.
  </Card>

  <Card title="Codex CLI" icon="https://mintcdn.com/qyhfrankgmailcom/--B4GB7WTjd2TEz0/images/common/clients/logos/openai.svg?fit=max&auto=format&n=--B4GB7WTjd2TEz0&q=85&s=19f4b841fced4e7e03466456edf9bea6" href="/en/clients/codex" width="721" height="721" data-path="images/common/clients/logos/openai.svg">
    Uses OpenAI Responses. Configure a separate `aiohub` provider.
  </Card>

  <Card title="OpenCode" icon="https://mintcdn.com/qyhfrankgmailcom/--B4GB7WTjd2TEz0/images/common/clients/logos/opencode-logo.svg?fit=max&auto=format&n=--B4GB7WTjd2TEz0&q=85&s=e9644f7bb1c2d12342347a19656f82b1" href="/en/clients/opencode" width="240" height="300" data-path="images/common/clients/logos/opencode-logo.svg">
    Configure an OpenCode provider and choose Chat Completions or Responses by model.
  </Card>

  <Card title="Gemini CLI" icon="https://mintcdn.com/qyhfrankgmailcom/--B4GB7WTjd2TEz0/images/common/clients/logos/gemini.svg?fit=max&auto=format&n=--B4GB7WTjd2TEz0&q=85&s=74855df7a8ecbc7faffc7175094fe3d2" href="/en/clients/gemini-cli" width="65" height="65" data-path="images/common/clients/logos/gemini.svg">
    Uses the Gemini API. Set `GOOGLE_GEMINI_BASE_URL` to the gateway root.
  </Card>

  <Card title="VS Code" icon="https://mintcdn.com/qyhfrankgmailcom/--B4GB7WTjd2TEz0/images/common/clients/logos/vscode.svg?fit=max&auto=format&n=--B4GB7WTjd2TEz0&q=85&s=f6255f57854fbbc576f36e32bdc609cf" href="/en/clients/vscode" width="100" height="100" data-path="images/common/clients/logos/vscode.svg">
    Use AIOHub in VS Code through the Claude Code and Codex extensions.
  </Card>

  <Card title="OpenAI-compatible apps" icon="plug" href="/en/clients/api-compatible">
    Use this for Cline, Cursor, Continue, Aider, Open WebUI, LibreChat, and similar apps.
  </Card>

  <Card title="Imagine" icon="image" href="/en/clients/imagine">
    Use Imagine for image generation and editing.
  </Card>

  <Card title="Cherry Studio" icon="message-square" href="/en/clients/cherry-studio">
    Desktop chat client. Fill the provider fields in the app.
  </Card>

  <Card title="LobeHub" icon="messages-square" href="/en/clients/lobehub">
    Use `https://chat.aiohub.org` and configure AIOHub in language model settings.
  </Card>

  <Card title="Claude Cowork 3P" icon="briefcase-business" href="/en/clients/claude-cowork-3p">
    Connect the Cowork and Code tabs in Claude Desktop to the AIOHub gateway.
  </Card>

  <Card title="Chatbox" icon="message-circle" href="/en/clients/chatbox">
    Use an OpenAI-compatible custom provider on mobile, desktop, or web.
  </Card>
</Columns>

## Writing SDK code or direct API calls

If you are writing code instead of configuring an existing app, start with [API overview](/en/api/overview) and [Authentication](/en/api/authentication). Client pages cover app fields and tool-specific setup.

## Common setup symptoms

| Symptom                                               | Check first                                                                 |
| ----------------------------------------------------- | --------------------------------------------------------------------------- |
| Error path includes `/v1/v1`                          | Base URL has an extra `/v1`                                                 |
| Error path repeats `/chat/completions`                | Base URL was set to a full endpoint                                         |
| Model not found                                       | The group assigned to the API key may not include that model                |
| Client cannot connect                                 | Check local environment variables, proxy settings, and client configuration |
| Service returns normally but the client shows failure | Check client response parsing and retry settings                            |