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

# SDKs & Libraries

> Recommended SDKs and libraries for TokenLab integrations

## Overview

TokenLab is easiest to integrate by matching the SDK to the behavior you need:

* official **Anthropic SDKs** for Claude-native `/v1/messages` behavior
* official **Gemini / Google AI** tooling for Gemini-native request shapes
* official **OpenAI SDKs** for OpenAI-compatible `/v1` routes and Responses-style usage

Use the integration pages for support boundaries. In this docs set, "recommended" or "supported" means a documented setup path exists; it does not automatically cover every framework helper or provider-specific feature.

## Recommended SDKs

<CardGroup cols={2}>
  <Card title="OpenAI Python" icon="python" href="https://github.com/openai/openai-python">
    `pip install openai`
  </Card>

  <Card title="OpenAI Node" icon="node-js" href="https://github.com/openai/openai-node">
    `npm install openai`
  </Card>

  <Card title="OpenAI Go" icon="golang" href="https://github.com/openai/openai-go">
    `go get github.com/openai/openai-go/v3`
  </Card>

  <Card title="Anthropic SDK" icon="robot" href="https://github.com/anthropics/anthropic-sdk-python">
    Native Claude Messages API support
  </Card>
</CardGroup>

## OpenAI SDK Example

Use this when you are migrating an existing OpenAI-compatible client or want the `/v1` compatibility path. For Claude-native or Gemini-native features, use the matching native SDK instead.

```python theme={null}
from openai import OpenAI

client = OpenAI(
    api_key="sk-your-api-key",
    base_url="https://api.tokenlab.sh/v1"
)

response = client.responses.create(
    model="gpt-5.4",
    input="Explain TokenLab in one sentence."
)

print(response.output_text)
```

## OpenAI Go Example

```go theme={null}
package main

import (
    openai "github.com/openai/openai-go/v3"
    "github.com/openai/openai-go/v3/option"
)

func main() {
    client := openai.NewClient(
        option.WithAPIKey("sk-your-api-key"),
        option.WithBaseURL("https://api.tokenlab.sh/v1"),
    )

    _ = client
}
```

## Anthropic SDK Example

```python theme={null}
from anthropic import Anthropic

client = Anthropic(
    api_key="sk-your-api-key",
    base_url="https://api.tokenlab.sh"
)

message = client.messages.create(
    model="claude-sonnet-4-6",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello!"}]
)
```

## Which SDK Should You Use?

| Goal                                   | Recommended path                    |
| -------------------------------------- | ----------------------------------- |
| Provider-native behavior               | Native provider SDK or route        |
| Existing OpenAI-compatible client      | OpenAI SDK on TokenLab `/v1`        |
| Portable chat / embeddings             | OpenAI-compatible `/v1` routes      |
| Claude-native features                 | Anthropic SDK                       |
| Gemini-native request shapes           | Gemini-native API / SDK             |
| LangChain / LlamaIndex / Vercel AI SDK | Use the dedicated integration pages |

## Best Practices

<AccordionGroup>
  <Accordion title="Choose native routes when behavior matters">
    Use Anthropic or Gemini native routes for provider-specific fields, tools, streaming details, and other behavior that should not be translated through another format.
  </Accordion>

  <Accordion title="Use OpenAI-compatible `/v1` for migration">
    OpenAI-compatible SDKs are the best path for existing OpenAI-style clients, portable chat, and embeddings. Treat them as a compatibility route, not as the only TokenLab API surface.
  </Accordion>

  <Accordion title="Keep base URLs explicit">
    Pass TokenLab base URLs directly in your client configuration instead of relying on older environment-variable aliases.
  </Accordion>
</AccordionGroup>