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

# 计算 Token 数量（Claude）

## 简介

计算 Claude 消息的 token 数量，用于在发送请求前预估成本。此端点**不消耗配额**，仅进行本地计算。

## 认证

<ParamField header="Authorization" type="string" required>
  Bearer Token，如 `Bearer sk-xxxxxxxxxx`
</ParamField>

## 请求参数

<ParamField body="model" type="string" required>
  Claude 模型标识，支持的模型包括：

  * `claude-opus-4-5-20251101`（推荐替代 claude-3-opus）
  * `claude-haiku-4-5-20251001`
  * `claude-sonnet-4-5-20250929`
  * `claude-sonnet-4-20250514`
  * 其他 Claude 系列模型
</ParamField>

<ParamField body="messages" type="array" required>
  对话消息列表，每个元素包含 `role`（user/assistant）和 `content`。`content` 可以是字符串或媒体内容数组。

  支持的内容类型：

  * 纯文本消息
  * 多模态消息（包含图片）
  * 工具调用结果
</ParamField>

<ParamField body="system" type="string|array">
  系统提示词（可选），可以是字符串或媒体内容数组。用于设定模型的行为和角色。
</ParamField>

<ParamField body="tools" type="array">
  工具定义列表（可选），用于计算工具调用相关的 token 数量。
</ParamField>

## 响应参数

<ResponseField name="input_tokens" type="number">
  输入消息的总 token 数量，包括：

  * system prompt 的 token 数
  * 所有 messages 的 token 数
  * tools 定义的 token 数（如果有）
</ResponseField>

## 基础示例

<Tabs>
  <Tab title="简单文本消息">
    ```bash theme={null}
    curl -X POST "https://api.leapx-hub.com/v1/messages/count_tokens" \
      -H "Content-Type: application/json" \
      -H "Authorization: Bearer sk-xxxxxxxxxx" \
      -d '{
        "model": "claude-sonnet-4-5-20250929",
        "messages": [
          {
            "role": "user",
            "content": "Hello, how are you?"
          }
        ]
      }'
    ```
  </Tab>

  <Tab title="带 system prompt">
    ```bash theme={null}
    curl -X POST "https://api.leapx-hub.com/v1/messages/count_tokens" \
      -H "Content-Type: application/json" \
      -H "Authorization: Bearer sk-xxxxxxxxxx" \
      -d '{
        "model": "claude-sonnet-4-5-20250929",
        "system": "You are a helpful AI assistant.",
        "messages": [
          {
            "role": "user",
            "content": "What is artificial intelligence?"
          }
        ]
      }'
    ```
  </Tab>

  <Tab title="多轮对话">
    ```bash theme={null}
    curl -X POST "https://api.leapx-hub.com/v1/messages/count_tokens" \
      -H "Content-Type: application/json" \
      -H "Authorization: Bearer sk-xxxxxxxxxx" \
      -d '{
        "model": "claude-sonnet-4-5-20250929",
        "messages": [
          {
            "role": "user",
            "content": "你好"
          },
          {
            "role": "assistant",
            "content": "你好！有什么我可以帮助你的吗？"
          },
          {
            "role": "user",
            "content": "给我讲讲人工智能的历史"
          }
        ]
      }'
    ```
  </Tab>

  <Tab title="Python 示例">
    ```python theme={null}
    from anthropic import Anthropic

    client = Anthropic(
        api_key="sk-xxxxxxxxxx",
        base_url="https://api.leapx-hub.com"
    )

    # 计算 token 数量
    response = client.messages.count_tokens(
        model="claude-sonnet-4-5-20250929",
        system="You are a helpful assistant.",
        messages=[
            {"role": "user", "content": "Hello, Claude!"}
        ]
    )

    print(f"Input tokens: {response.input_tokens}")
    ```
  </Tab>
</Tabs>

<ResponseExample>
  ```json theme={null}
  {
    "input_tokens": 14
  }
  ```
</ResponseExample>

## 高级用例

### 带工具定义的计算

```bash theme={null}
curl -X POST "https://api.leapx-hub.com/v1/messages/count_tokens" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-xxxxxxxxxx" \
  -d '{
    "model": "claude-sonnet-4-5-20250929",
    "messages": [
      {
        "role": "user",
        "content": "What is the weather in San Francisco?"
      }
    ],
    "tools": [
      {
        "name": "get_weather",
        "description": "Get the current weather in a given location",
        "input_schema": {
          "type": "object",
          "properties": {
            "location": {
              "type": "string",
              "description": "The city and state, e.g. San Francisco, CA"
            }
          },
          "required": ["location"]
        }
      }
    ]
  }'
```

### 多模态内容计算

```bash theme={null}
curl -X POST "https://api.leapx-hub.com/v1/messages/count_tokens" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-xxxxxxxxxx" \
  -d '{
    "model": "claude-sonnet-4-5-20250929",
    "messages": [
      {
        "role": "user",
        "content": [
          {
            "type": "text",
            "text": "What is in this image?"
          },
          {
            "type": "image",
            "source": {
              "type": "url",
              "url": "https://example.com/image.jpg"
            }
          }
        ]
      }
    ]
  }'
```

## 使用场景

### 1. 成本预估

在发送大量请求前，先计算 token 数量以预估成本：

```python theme={null}
# 批量计算成本
messages_batch = [...]  # 批量消息
total_tokens = 0

for messages in messages_batch:
    response = client.messages.count_tokens(
        model="claude-sonnet-4-5-20250929",
        messages=messages
    )
    total_tokens += response.input_tokens

# 根据定价计算总成本
cost = total_tokens * price_per_token
print(f"预估成本: ${cost:.4f}")
```

### 2. 上下文窗口管理

检查消息是否超过模型的上下文窗口限制：

```python theme={null}
MAX_CONTEXT_WINDOW = 200000  # Claude Sonnet 4.5 的上下文窗口

response = client.messages.count_tokens(
    model="claude-sonnet-4-5-20250929",
    messages=long_conversation
)

if response.input_tokens > MAX_CONTEXT_WINDOW:
    print(f"警告：消息 token 数 ({response.input_tokens}) 超过上下文窗口限制")
    # 执行消息截断或摘要
```

### 3. 优化提示词

比较不同提示词的 token 消耗：

```python theme={null}
prompts = [
    "简洁版提示词...",
    "详细版提示词...",
    "超详细版提示词..."
]

for prompt in prompts:
    response = client.messages.count_tokens(
        model="claude-sonnet-4-5-20250929",
        system=prompt,
        messages=[{"role": "user", "content": "测试"}]
    )
    print(f"{len(prompt)} 字符 -> {response.input_tokens} tokens")
```

## 注意事项

<Warning>
  * 图片 token 使用固定估算值（约 1000 tokens），实际可能因分辨率不同而变化
  * 不包括 `max_tokens` 等输出相关参数，仅计算输入 token 数
  * 此端点不会发起实际的 AI 请求，不消耗配额
</Warning>

## 错误处理

### 缺少必需参数

```json theme={null}
{
  "type": "error",
  "error": {
    "type": "invalid_request_error",
    "message": "Key: 'ClaudeCountTokensRequest.Model' Error:Field validation for 'Model' failed on the 'required' tag"
  }
}
```

### 无效的 API Key

```json theme={null}
{
  "error": {
    "message": "无效的 token",
    "type": "invalid_request_error"
  }
}
```

## 相关资源

* [创建消息请求（Claude）](/cn/api-reference/endpoint/messages) - 发送实际的 Claude 请求
* [模型列表](/cn/api-reference/models) - 查看支持的 Claude 模型
* [定价说明](https://leapx-hub.com) - 了解 token 计费标准