DeepSeek API 使用手册

一、概述

DeepSeek API 完全兼容 OpenAI API 格式，现有 OpenAI SDK 和代码可零成本迁移。同时提供 Anthropic API 兼容格式。DeepSeek 以极低价格提供顶级模型能力，性价比业界领先。

二、认证

Authorization: Bearer your-deepseek-api-key

API Key 在 platform.deepseek.com/api_keys 创建。

三、Base URL

OpenAI兼容格式：https://api.deepseek.com
Anthropic兼容格式：https://api.deepseek.com/anthropic
聊天补全端点：https://api.deepseek.com/chat/completions

四、模型列表

五、使用 OpenAI SDK 调用

from openai import OpenAI

client = OpenAI(
    api_key="your-deepseek-api-key",
    base_url="https://api.deepseek.com"
)

# 非思考模式
response = client.chat.completions.create(
    model="deepseek-v4-pro",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Hello!"}
    ],
    stream=False
)
print(response.choices[0].message.content)

六、思考模式

DeepSeek 独有的深度推理能力，模型在回答前进行长链推理：

response = client.chat.completions.create(
    model="deepseek-v4-pro",
    messages=[
        {"role": "user", "content": "求解 x^2 + 5x + 6 = 0"}
    ],
    thinking={"type": "enabled"},
    reasoning_effort="high",  # low / medium / high
    stream=False
)

思考模式响应结构

{
  "choices": [{
    "message": {
      "role": "assistant",
      "content": "方程 x^2 + 5x + 6 = 0 的解为 x = -2 或 x = -3",
      "reasoning_content": "这是一个一元二次方程...n使用因式分解...n(x+2)(x+3)=0..."
    }
  }]
}

reasoning_effort 参数

七、流式输出

stream = client.chat.completions.create(
    model="deepseek-v4-pro",
    messages=[{"role": "user", "content": "写一首诗"}],
    stream=True
)

for chunk in stream:
    # 普通内容
    if chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="")
    # 思考内容（思考模式下）
    if hasattr(chunk.choices[0].delta, 'reasoning_content') and chunk.choices[0].delta.reasoning_content:
        # 可选择不输出思考过程
        pass

八、cURL 调用示例

curl https://api.deepseek.com/chat/completions 
  -H "Content-Type: application/json" 
  -H "Authorization: Bearer $DEEPSEEK_API_KEY" 
  -d '{
    "model": "deepseek-v4-pro",
    "messages": [
      {"role": "system", "content": "You are a helpful assistant."},
      {"role": "user", "content": "Hello!"}
    ],
    "thinking": {"type": "enabled"},
    "reasoning_effort": "high",
    "stream": false
  }'

九、Function Calling

与 OpenAI 格式完全一致：

tools = [{
    "type": "function",
    "function": {
        "name": "get_weather",
        "description": "获取天气信息",
        "parameters": {
            "type": "object",
            "properties": {
                "city": {"type": "string", "description": "城市名"}
            },
            "required": ["city"]
        }
    }
}]

response = client.chat.completions.create(
    model="deepseek-v4-pro",
    messages=[{"role": "user", "content": "北京天气如何？"}],
    tools=tools
)

十、JSON 输出

response = client.chat.completions.create(
    model="deepseek-v4-pro",
    messages=[{"role": "user", "content": "列出中国前5大城市"}],
    response_format={"type": "json_object"}
)

十一、定价

十二、Agent 工具集成

无需编码，直接在主流AI工具中使用：

十三、速率限制

默认限制（付费用户）：

• RPM: 60（v4-pro）/ 120（v4-flash）
• TPM: 2,000,000

余额不足时自动降级为免费层级限速。

十四、错误码