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

# Client integrations overview

> Connect OpenPAI in common AI apps / IDEs / SDKs

As long as an app supports the OpenAI-compatible **custom `base_url` + `API key`** setting, you can switch it to OpenPAI in one step.

## Common configuration

| Item     | Value                                                |
| -------- | ---------------------------------------------------- |
| Base URL | `https://openp.ai/v1`                                |
| API key  | `sk-XXXXXXXXXXXXXXXX` (from the console Tokens page) |
| Protocol | OpenAI / Anthropic / Google, your choice             |

## Desktop GUI clients

<CardGroup cols={2}>
  <Card title="Cherry Studio" icon="cherries" href="/en/integrations/cherry-studio">
    Cross-platform multi-model chat client, recommended for getting started.
  </Card>

  <Card title="LobeChat" icon="comments" href="/en/integrations/lobe-chat">
    Open-source, self-hostable modern chat UI.
  </Card>

  <Card title="NextChat" icon="message" href="/en/integrations/nextchat">
    Lightweight chat UI with one-click Vercel deployment.
  </Card>

  <Card title="ChatBox" icon="window-restore" href="/en/integrations/chatbox">
    Desktop ChatGPT client for Windows/Mac/Linux.
  </Card>
</CardGroup>

## Coding assistants / IDEs

<CardGroup cols={2}>
  <Card title="Cursor" icon="code" href="/en/integrations/cursor">
    AI-first code editor, supports a custom OpenAI base URL.
  </Card>

  <Card title="Cline / Continue" icon="terminal" href="/en/integrations/cline">
    Open-source AI coding agents in VS Code.
  </Card>
</CardGroup>

## Application frameworks

<CardGroup cols={2}>
  <Card title="Dify" icon="cubes" href="/en/integrations/dify">
    LLMOps platform, using OpenPAI as a model provider.
  </Card>

  <Card title="LangChain / LlamaIndex" icon="link">
    Use the official OpenAI / Anthropic provider directly — just change base\_url.
  </Card>
</CardGroup>

## Official SDKs

<CardGroup cols={3}>
  <Card title="OpenAI SDK" icon="o" href="/en/integrations/openai-sdk">
    Python / Node / Go / Java
  </Card>

  <Card title="Anthropic SDK" icon="a" href="/en/integrations/anthropic-sdk">
    Python / Node, using `/v1/messages`
  </Card>

  <Card title="google-genai" icon="g" href="/en/integrations/google-sdk">
    Python / JS, using the Gemini native protocol
  </Card>
</CardGroup>

## Plain cURL

<Card title="cURL cheatsheet" icon="terminal" href="/en/integrations/curl-examples">
  cURL templates for common tasks: chat, streaming, vision, tools, embeddings, images.
</Card>

## General tips

<AccordionGroup>
  <Accordion title="What if I can't find a model ID?">
    Manually enter the model ID shown in the OpenPAI console in the client's **model** settings.
    Some clients hard-code OpenAI models in the dropdown — switch to **custom model** mode.
  </Accordion>

  <Accordion title="Some apps require base_url without `/v1`">
    OpenPAI accepts requests both with and without `/v1`.

    * OpenAI SDK-style clients: `https://openp.ai/v1`
    * Anthropic / Google clients: `https://openp.ai`
  </Accordion>

  <Accordion title="Do I need CORS / direct browser calls?">
    OpenPAI enables CORS by default, allowing direct browser-side calls.
    But it is **strongly discouraged** to expose your API key in the frontend — forward through your own backend instead.
  </Accordion>
</AccordionGroup>