跳转至

Deepseek reasoning 对话格式(类Chat Completions)

📝 简介

Deepseek-reasoner 是 DeepSeek 推出的推理模型。在输出最终回答之前,模型会先输出一段思维链内容,以提升最终答案的准确性。API 向用户开放 deepseek-reasoner 思维链的内容,以供用户查看、展示、蒸馏使用。

💡 请求示例

基础文本对话 ✅

curl https://api.deepseek.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $NEWAPI_API_KEY" \
  -d '{
    "model": "deepseek-reasoner",
    "messages": [
      {
        "role": "user",
        "content": "9.11 and 9.8, which is greater?"
      }
    ],
    "max_tokens": 4096
  }'

响应示例:

{
  "id": "chatcmpl-123",
  "object": "chat.completion",
  "created": 1677652288,
  "model": "deepseek-reasoner",
  "choices": [{
    "index": 0,
    "message": {
      "role": "assistant",
      "reasoning_content": "让我一步步思考:\n1. 我们需要比较9.11和9.8的大小\n2. 两个数都是小数,我们可以直接比较\n3. 9.8 = 9.80\n4. 9.11 < 9.80\n5. 所以9.8更大",
      "content": "9.8 is greater than 9.11."
    },
    "finish_reason": "stop"
  }],
  "usage": {
    "prompt_tokens": 10,
    "completion_tokens": 15,
    "total_tokens": 25
  }
}

流式响应 ✅

curl https://api.deepseek.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $NEWAPI_API_KEY" \
  -d '{
    "model": "deepseek-reasoner",
    "messages": [
      {
        "role": "user",
        "content": "9.11 and 9.8, which is greater?"
      }
    ],
    "stream": true
  }'

流式响应示例:

{"id":"chatcmpl-123","object":"chat.completion.chunk","created":1694268190,"model":"deepseek-reasoner","choices":[{"index":0,"delta":{"role":"assistant","reasoning_content":"让我"},"finish_reason":null}]}

{"id":"chatcmpl-123","object":"chat.completion.chunk","created":1694268190,"model":"deepseek-reasoner","choices":[{"index":0,"delta":{"reasoning_content":"一步步"},"finish_reason":null}]}

{"id":"chatcmpl-123","object":"chat.completion.chunk","created":1694268190,"model":"deepseek-reasoner","choices":[{"index":0,"delta":{"reasoning_content":"思考:"},"finish_reason":null}]}

// ... 更多思维链内容 ...

{"id":"chatcmpl-123","object":"chat.completion.chunk","created":1694268190,"model":"deepseek-reasoner","choices":[{"index":0,"delta":{"content":"9.8"},"finish_reason":null}]}

{"id":"chatcmpl-123","object":"chat.completion.chunk","created":1694268190,"model":"deepseek-reasoner","choices":[{"index":0,"delta":{"content":" is greater"},"finish_reason":null}]}

// ... 更多最终答案内容 ...

{"id":"chatcmpl-123","object":"chat.completion.chunk","created":1694268190,"model":"deepseek-reasoner","choices":[{"index":0,"delta":{},"finish_reason":"stop"}]}

📮 请求

端点

POST /v1/chat/completions

鉴权方法

在请求头中包含以下内容进行 API 密钥认证:

Authorization: Bearer $NEWAPI_API_KEY

其中 $DEEPSEEK_API_KEY 是您的 API 密钥。

请求体参数

messages

  • 类型:数组
  • 必需:是

到目前为止包含对话的消息列表。请注意,如果您在输入的 messages 序列中传入了 reasoning_content,API 会返回 400 错误。

model

  • 类型:字符串
  • 必需:是
  • 值:deepseek-reasoner

要使用的模型 ID。目前仅支持 deepseek-reasoner。

max_tokens

  • 类型:整数
  • 必需:否
  • 默认值:4096
  • 最大值:8192

最终回答的最大长度(不含思维链输出)。请注意,思维链的输出最多可以达到 32K tokens。

stream

  • 类型:布尔值
  • 必需:否
  • 默认值:false

是否使用流式响应。

不支持的参数

以下参数当前不支持:

  • temperature
  • top_p
  • presence_penalty
  • frequency_penalty
  • logprobs
  • top_logprobs

注意:为了兼容已有软件,设置 temperature、top_p、presence_penalty、frequency_penalty 参数不会报错,但也不会生效。设置 logprobs、top_logprobs 会报错。

支持的功能

  • 对话补全
  • 对话前缀续写 (Beta)

不支持的功能

  • Function Call
  • Json Output
  • FIM 补全 (Beta)

📥 响应

成功响应

返回一个聊天补全对象,如果请求被流式传输,则返回聊天补全块对象的流式序列。

id

  • 类型:字符串
  • 说明:响应的唯一标识符

object

  • 类型:字符串
  • 说明:对象类型,值为 "chat.completion"

created

  • 类型:整数
  • 说明:响应创建时间戳

model

  • 类型:字符串
  • 说明:使用的模型名称,值为 "deepseek-reasoner"

choices

  • 类型:数组
  • 说明:包含生成的回复选项
  • 属性:
  • index: 选项索引
  • message: 包含角色、思维链内容和最终回答的消息对象
    • role: 角色,值为 "assistant"
    • reasoning_content: 思维链内容
    • content: 最终回答内容
  • finish_reason: 完成原因

usage

  • 类型:对象
  • 说明:token 使用统计
  • 属性:
  • prompt_tokens: 提示使用的 token 数
  • completion_tokens: 补全使用的 token 数
  • total_tokens: 总 token 数

📝 上下文拼接说明

在每一轮对话过程中,模型会输出思维链内容(reasoning_content)和最终回答(content)。在下一轮对话中,之前轮输出的思维链内容不会被拼接到上下文中,如下图所示:

Deepseek reasoning 上下文拼接示意图

注意

如果您在输入的 messages 序列中,传入了reasoning_content,API 会返回 400 错误。因此,请删除 API 响应中的 reasoning_content 字段,再发起 API 请求,方法如下方使用示例所示。

使用示例:

from openai import OpenAI
client = OpenAI(api_key="<DeepSeek API Key>", base_url="https://api.deepseek.com")

# 第一轮对话
messages = [{"role": "user", "content": "9.11 and 9.8, which is greater?"}]
response = client.chat.completions.create(
    model="deepseek-reasoner",
    messages=messages
)

reasoning_content = response.choices[0].message.reasoning_content
content = response.choices[0].message.content

# 第二轮对话 - 只拼接最终回答content
messages.append({'role': 'assistant', 'content': content})
messages.append({'role': 'user', 'content': "How many Rs are there in the word 'strawberry'?"})
response = client.chat.completions.create(
    model="deepseek-reasoner", 
    messages=messages
)

流式响应示例:

# 第一轮对话
messages = [{"role": "user", "content": "9.11 and 9.8, which is greater?"}]
response = client.chat.completions.create(
    model="deepseek-reasoner",
    messages=messages,
    stream=True
)

reasoning_content = ""
content = ""

for chunk in response:
    if chunk.choices[0].delta.reasoning_content:
        reasoning_content += chunk.choices[0].delta.reasoning_content
    else:
        content += chunk.choices[0].delta.content

# 第二轮对话 - 只拼接最终回答content
messages.append({"role": "assistant", "content": content})
messages.append({'role': 'user', 'content': "How many Rs are there in the word 'strawberry'?"})
response = client.chat.completions.create(
    model="deepseek-reasoner",
    messages=messages,
    stream=True
)