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

# Long context

# 1M Token Long Context Window

Anthropic’s Claude model family supports expanding the context window from the default 200K tokens to **1,000,000 tokens (1M)**—5× the default capacity. With ShuYou, you can enable this capability effortlessly to handle large-scale document analysis, code review, long-running conversations, and more.

<Tip title="💡 Key Benefits">
  * **Massive context**: Process roughly 750,000 English words or thousands of pages in a single request
  * **Deep analysis**: Ideal for large codebase reviews, long-form literature analysis, full conversation history, and more
  * **Seamless integration**: Enable it via ShuYou’s unified interface by adding a single request header
</Tip>

## Supported Models

The following Claude models currently support the 1M-token context window:

| Model                 | Default Context Window | Extended Context Window |
| --------------------- | ---------------------- | ----------------------- |
| **Claude Opus 4.6**   | 200K tokens            | 1M tokens               |
| **Claude Sonnet 4.5** | 200K tokens            | 1M tokens               |
| **Claude Sonnet 4**   | 200K tokens            | 1M tokens               |

<Warning title="Note">
  The 1M-token context window is currently a Beta feature, and functionality and pricing may change in future versions.
</Warning>

## How to Enable

To use the 1M-token context window, add the request header `anthropic-beta: context-1m-2025-08-07`. If you don’t include this header, the model will use the default 200K-token context window.

### Method 1: Use the OpenAI-Compatible API (Recommended)

```python title="Python" theme={null}
from openai import OpenAI

client = OpenAI(
    base_url="https://api.shuyou.ai/v1",
    api_key="<your ShuYou_API_KEY>", # [!code highlight]
)

response = client.chat.completions.create(
    model="anthropic/claude-sonnet-4.5",
    messages=[
        {
            "role": "user",
            "content": "Please analyze the core content of the following long document..."
        }
    ],
    extra_headers={
        "anthropic-beta": "context-1m-2025-08-07" # [!code highlight]
    }
)

print(response.choices[0].message.content)
```

```ts title="TypeScript" theme={null}
import OpenAI from "openai";

const openai = new OpenAI({
  baseURL: "https://api.shuyou.ai/v1",
  apiKey: "<your ShuYou_API_KEY>", // [!code highlight]
});

async function main() {
  const response = await openai.chat.completions.create(
    {
      model: "anthropic/claude-sonnet-4.5",
      messages: [
        {
          role: "user",
          content: "Please analyze the core content of the following long document...",
        },
      ],
    },
    {
      headers: {
        "anthropic-beta": "context-1m-2025-08-07", // [!code highlight]
      },
    },
  );

  console.log(response.choices[0].message.content);
}

main();
```

```bash title="Shell (cURL)" theme={null}
curl "https://api.shuyou.ai/v1/chat/completions" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $ShuYou_API_KEY" \
  -H "anthropic-beta: context-1m-2025-08-07" \
  -d '{
    "model": "anthropic/claude-sonnet-4.5",
    "messages": [
      {
        "role": "user",
        "content": "Please analyze the core content of the following long document..."
      }
    ]
  }'
```

***

### Method 2: Use the Native Anthropic API

```python title="Python" theme={null}
import anthropic

client = anthropic.Anthropic(
    base_url="https://api.shuyou.ai",
    api_key="<your ShuYou_API_KEY>", # [!code highlight]
)

response = client.beta.messages.create(
    model="anthropic/claude-sonnet-4.5",
    betas=["context-1m-2025-08-07"], # [!code highlight]
    max_tokens=4096,
    messages=[
        {
            "role": "user",
            "content": "Please analyze the core content of the following long document..."
        }
    ]
)

print(response.content[0].text)
```

```ts title="TypeScript" theme={null}
import Anthropic from "@anthropic-ai/sdk";

const client = new Anthropic({
  baseURL: "https://api.shuyou.ai",
  apiKey: "<your ShuYou_API_KEY>", // [!code highlight]
});

async function main() {
  const response = await client.beta.messages.create({
    model: "anthropic/claude-sonnet-4.5",
    betas: ["context-1m-2025-08-07"], // [!code highlight]
    max_tokens: 4096,
    messages: [
      {
        role: "user",
        content: "Please analyze the core content of the following long document...",
      },
    ],
  });

  console.log(response.content[0].text);
}

main();
```

```bash title="Shell (cURL)" theme={null}
curl "https://api.shuyou.ai/v1/messages" \
  -H "Content-Type: application/json" \
  -H "x-api-key: $ShuYou_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "anthropic-beta: context-1m-2025-08-07" \
  -d '{
    "model": "anthropic/claude-sonnet-4.5",
    "max_tokens": 4096,
    "messages": [
      {
        "role": "user",
        "content": "Please analyze the core content of the following long document..."
      }
    ]
  }'
```

## Long-Context Pricing

When the number of tokens in a request exceeds 200K, long-context pricing will apply automatically. The multipliers are as follows:

| Pricing Item      | Up to 200K         | Over 200K                 |
| ----------------- | ------------------ | ------------------------- |
| **Input tokens**  | 1x (standard rate) | 2x (double rate)          |
| **Output tokens** | 1x (standard rate) | 1.5x (1.5× standard rate) |

<Tip title="💰 Billing Notes">
  * Only the portion that **actually exceeds 200K** is billed at the higher multiplier; the portion within 200K is still billed at the standard rate
  * The output-token multiplier (1.5x) is lower than the input-token multiplier (2x), keeping the overall cost increase manageable
  * We recommend using this together with [Prompt Caching](/en/guide/advanced/prompt-cache) to significantly reduce costs in long-context scenarios
</Tip>

## Best Practices

### Recommended Use Cases

<Tip title="💡 Scenarios where 1M context is recommended">
  * **Large-scale code review**: Load an entire repository at once for analysis and refactoring suggestions
  * **Long-document analysis**: Process full legal contracts, academic papers, technical specifications, and more
  * **Multi-document comparison**: Analyze multiple related documents simultaneously for cross-referencing and comparison
  * **Full conversation history**: Preserve complete context in very long conversations to avoid losing information
  * **Data analysis**: Process large volumes of structured or unstructured data
</Tip>

### Optimization Tips

<Tip title="💡 Improve long-context request quality">
  1. **Put critical information first**: Place the most important content at the beginning of the prompt—models pay the most attention to the start
  2. **Combine with prompt caching**: For long documents reused repeatedly, cache static content via `cache_control` to save up to 90% of input costs
  3. **Control context size wisely**: You don’t need to use the full 1M window every time—load only what’s relevant to the current task
  4. **Use structured markup**: Organize long documents with clear XML tags or Markdown headings to help the model locate key information
</Tip>

## FAQ

### What happens if I don’t add the Beta header?

If you don’t add the `anthropic-beta: context-1m-2025-08-07` header, the model uses the default 200K-token context window. Any content beyond 200K cannot be processed.

### Is the 1M context window stable and production-ready?

This feature is currently in Beta. While it can already be used in production, Anthropic may adjust feature details and pricing in the future.

### Can I use it together with other features?

Yes. The 1M context window can be used together with features such as [Prompt Caching](/en/guide/advanced/prompt-cache), [Tool Calling](/en/guide/advanced/tool-calls), and [Streaming](/en/guide/advanced/streaming). Prompt caching is especially recommended to optimize costs for long-context workloads.