火山方舟
火山方舟
文档指南
请输入
调用
视频生成
工具调用
使用 Responses API
推理
训练
模型精调
Managed Agents
配置 Agent 环境
管理 Agent 上下文
高级能力
- 文档首页
火山方舟调用进阶使用上下文管理
复制全文
下载 pdf上下文管理
文档反馈
问问助手模型上下文（Context）包括输入信息（问题）和输出信息（回答），当触发深度思考时输出还包括思维链内容（CoT）。思维链内容展现模型处理问题的过程，包括将问题拆分为多个问题进行处理，生成多种回复综合得出更好回答的过程。在多轮对话中，上下文管理在保持话题一致性等方面非常重要，对此方舟提供了一些管理上下文的方法。
上下文输入
手动输入上下文
Curl
Python
Go
Java
OpenAI SDK
curl https://ark.cn-beijing.volces.com/api/v3/chat/completions \
-H "Content-Type: application/json" \
-H "Authorization: Bearer $ARK_API_KEY" \
-d '{
"model": "doubao-seed-2-1-pro-260915",
"messages": [
{"role": "user", "content": "深度思考模型与非深度思考模型区别"},
{"role": "assistant", "content": "推理模型主要依靠逻辑、规则或概率等进行分析、推导和判断以得出结论或决策，非推理模型则是通过模式识别、统计分析或模拟等方式来实现数据描述、分类、聚类或生成等任务而不依赖显式逻辑推理。"},
{"role": "user", "content": "我要研究深度思考模型与非深度思考模型区别的课题，怎么体现我的专业性"}
]
}'
- 您可按需替换 Model ID。Model ID 查询见 模型列表。
import os
# Install SDK: python -m pip install --upgrade arkruntime
from arkruntime import Ark
client = Ark(
# The base URL for model invocation
base_url="https://ark.cn-beijing.volces.com/api/v3",
# Get API Key: https://ark.volcengine.com/region:cn-beijing/apikey
api_key=os.getenv('ARK_API_KEY'),
# Deep thinking takes longer; set a larger timeout, with 1,800 seconds or more recommended
timeout=1800,
)
# 创建一个对话请求
completion = client.chat.completions.create(
# Replace with Model ID
model = "doubao-seed-2-1-pro-260915",
messages=[
{"role": "user", "content": "研究深度思考模型与非深度思考模型区别"},
{"role": "assistant", "content": "推理模型主要依靠逻辑、规则或概率等进行分析、推导和判断以得出结论或决策，非推理模型则是通过模式识别、统计分析或模拟等方式来实现数据描述、分类、聚类或生成等任务而不依赖显式逻辑推理。"},
{"role": "user", "content": "我要研究深度思考模型与非深度思考模型区别的课题，怎么体现我的专业性"},
],
)
if hasattr(completion.choices[0].message, 'reasoning_content'):
print(completion.choices[0].message.reasoning_content)
print(completion.choices[0].message.content)
package main
import (
"context"
"fmt"
"os"
"time"
"github.com/volcengine/ark-runtime-go/arkruntime"
"github.com/volcengine/ark-runtime-go/arkruntime/model/chat"
)
func main() {
client := arkruntime.NewClientWithApiKey(
//通过 os.Getenv 从环境变量中获取 ARK_API_KEY
os.Getenv("ARK_API_KEY"),
// The base URL for model invocation
arkruntime.WithBaseUrl("https://ark.cn-beijing.volces.com/api/v3"),
//深度思考耗时更长，请设置更大的超时限制，推荐为30分钟及以上
arkruntime.WithTimeout(30*time.Minute),
)
// 创建一个上下文，通常用于传递请求的上下文信息，如超时、取消等
ctx := context.Background()
// 构建聊天完成请求，设置请求的模型和消息内容
req := &chat.ChatCompletionRequest{
// Replace with Model ID
Model: "doubao-seed-2-1-pro-260915",
Messages: []chat.ChatCompletionRequestMessage{
userMessage("研究深度思考模型与非深度思考模型区别"),
assistantMessage("推理模型主要依靠逻辑、规则或概率等进行分析、推导和判断以得出结论或决策，非推理模型则是通过模式识别、统计分析或模拟等方式来实现数据描述、分类、聚类或生成等任务而不依赖显式逻辑推理。"),
userMessage("我要研究深度思考模型与非深度思考模型区别的课题，怎么体现我的专业性"),
},
}
// 发送聊天完成请求，并将结果存储在 resp 中，将可能出现的错误存储在 err 中
resp, err := client.CreateChatCompletion(ctx, req)
if err != nil {
// 若出现错误，打印错误信息并终止程序
fmt.Printf("standard chat error: %v\n", err)
return
}
// 检查是否触发深度思考，触发则打印思维链内容
if resp.Choices[0].Message.ReasoningContent.IsSet() {
fmt.Println(resp.Choices[0].Message.ReasoningContent.Or(""))
}
// 打印聊天完成请求的响应结果
fmt.Println(resp.Choices[0].Message.Content)
}
func assistantMessage(content string) chat.ChatCompletionRequestMessage {
return chat.ChatCompletionRequestMessage{
OneOf: chat.NewChatCompletionRequestAssistantMessageChatCompletionRequestMessageSum(
chat.ChatCompletionRequestAssistantMessage{
Role: chat.ChatCompletionRequestAssistantMessageRoleAssistant,
Content: chat.NewOptNilChatCompletionMessageContent(chat.NewStringChatCompletionMessageContent(content)),
},
),
}
}
func userMessage(content string) chat.ChatCompletionRequestMessage {
return chat.ChatCompletionRequestMessage{
OneOf: chat.NewChatCompletionRequestUserMessageChatCompletionRequestMessageSum(
chat.ChatCompletionRequestUserMessage{
Role: chat.ChatCompletionRequestUserMessageRoleUser,
Content: chat.NewOptNilChatCompletionMessageContent(chat.NewStringChatCompletionMessageContent(content)),
},
),
}
}
package com.volcengine.ark.runtime;
import com.volcengine.ark.runtime.models.chat.*;
import com.volcengine.ark.runtime.service.ArkService;
import java.time.Duration;
import java.util.Arrays;
import java.util.List;
/**
* 这是一个示例类，展示了如何使用ArkService来完成聊天功能。
*/
public class ChatCompletionsExample {
public static void main(String[] args) {
// 从环境变量中获取API密钥
String apiKey = System.getenv("ARK_API_KEY");
// 创建ArkService实例
ArkService arkService = ArkService.builder()
.apiKey(apiKey)
.timeout(Duration.ofMinutes(30))// 深度思考耗时更长，请设置更大的超时限制，推荐为30分钟及以上
// The base URL for model invocation
.baseUrl("https://ark.cn-beijing.volces.com/api/v3")
.build();
// 多轮消息列表
final List<ChatCompletionRequestMessage> messages = Arrays.asList(
ChatCompletionRequestUserMessage.builder().role(ChatCompletionRequestMessageType.USER).content(ChatCompletionMessageContent.ofString("研究深度思考模型与非深度思考模型区别")).build(),
ChatCompletionRequestAssistantMessage.builder().role(ChatCompletionRequestMessageType.ASSISTANT).content(ChatCompletionMessageContent.ofString("推理模型主要依靠逻辑、规则或概率等进行分析、推导和判断以得出结论或决策，非推理模型则是通过模式识别、统计分析或模拟等方式来实现数据描述、分类、聚类或生成等任务而不依赖显式逻辑推理。")).build(),
ChatCompletionRequestUserMessage.builder().role(ChatCompletionRequestMessageType.USER).content(ChatCompletionMessageContent.ofString("我要研究深度思考模型与非深度思考模型区别的课题，怎么体现我的专业性")).build()
);
// 创建聊天完成请求
ChatCompletionRequest chatCompletionRequest = ChatCompletionRequest.builder()
.model("doubao-seed-2-1-pro-260915")//Replace with Model ID
.messages(messages) // 设置消息列表
.build();
// 发送聊天完成请求并打印响应
try {
// 获取响应并打印每个选择的消息内容
arkService.createChatCompletion(chatCompletionRequest)
.getChoices()
.forEach(choice -> {
// 进行空值检查后打印推理内容
if (choice.getMessage().getReasoningContent() != null) {
System.out.println("推理内容: " + choice.getMessage().getReasoningContent());
} else {
System.out.println("推理内容为空");
}
// 打印消息内容
System.out.println("消息内容: " + choice.getMessage().getContent());
});
} catch (Exception e) {
System.out.println("请求失败: " + e.getMessage());
} finally {
// 关闭服务执行器
arkService.shutdownExecutor();
}
}
}
import os
from openai import OpenAI
client = OpenAI(
# 从环境变量中读取方舟API Key
api_key=os.environ.get("ARK_API_KEY"),
base_url="https://ark.cn-beijing.volces.com/api/v3",
# 深度思考耗时更长，避免链接超时导致失败，请设置更大的超时限制，推荐为1800 秒及以上
timeout=1800,
)
completion = client.chat.completions.create(
# Replace with Model ID
model = "doubao-seed-2-1-pro-260915",
messages=[
{"role": "user", "content": "研究深度思考模型与非深度思考模型区别"},
{"role": "assistant", "content": "推理模型主要依靠逻辑、规则或概率等进行分析、推导和判断以得出结论或决策，非推理模型则是通过模式识别、统计分析或模拟等方式来实现数据描述、分类、聚类或生成等任务而不依赖显式逻辑推理。"},
{"role": "user", "content": "我要研究深度思考模型与非深度思考模型区别的课题，怎么体现我的专业性"},
],
)
if hasattr(completion.choices[0].message, 'reasoning_content'):
print(completion.choices[0].message.reasoning_content)
print(completion.choices[0].message.content)
通过 ID 输入上下文
Curl
Python
Go
Java
OpenAI SDK
curl https://ark.cn-beijing.volces.com/api/v3/responses \
-H "Authorization: Bearer $ARK_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "doubao-seed-2-1-pro-260915",
"input": "Hi，讲个笑话。"
}'
curl https://ark.cn-beijing.volces.com/api/v3/responses \
-H "Authorization: Bearer $ARK_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "doubao-seed-2-1-pro-260915",
"previous_response_id":"<id>",
"input": "这个笑话的笑点在哪？"
}'
第二次请求需将 curl 命令中的 <id> 替换为上一次请求返回的 Response id。
import os
from arkruntime import Ark
# Get API Key: https://ark.volcengine.com/region:cn-beijing/apikey
api_key = os.getenv('ARK_API_KEY')
client = Ark(
base_url='https://ark.cn-beijing.volces.com/api/v3',
api_key=api_key,
)
# Create the first-round conversation request
response = client.responses.create(
model="doubao-seed-2-1-pro-260915",
input="Hi，帮我讲个笑话。"
)
print(response)
# Create the second-round conversation request
second_response = client.responses.create(
model="doubao-seed-2-1-pro-260915",
previous_response_id=response.id,
input=[{"role": "user", "content": "这个笑话的笑点在哪？"}],
)
print(second_response)
package main
import (
"context"
"fmt"
"os"
"github.com/volcengine/ark-runtime-go/arkruntime"
"github.com/volcengine/ark-runtime-go/arkruntime/model/responses"
)
func main() {
client := arkruntime.NewClientWithApiKey(
// Get API Key: https://ark.volcengine.com/region:cn-beijing/apikey
os.Getenv("ARK_API_KEY"),
arkruntime.WithBaseUrl("https://ark.cn-beijing.volces.com/api/v3"),
)
ctx := context.Background()
// Create the first-round conversation request
resp, err := client.CreateResponses(ctx, &responses.ResponsesRequest{
Model: "doubao-seed-2-1-pro-260915",
Input: responses.NewStringResponsesInput("Hi，讲个笑话。"),
Store: responses.NewOptBool(true),
})
if err != nil {
fmt.Printf("response error: %v\n", err)
return
}
fmt.Println(resp)
id := resp.ID
inputMessage := easyMessage(responses.MessageRoleUser, "这个笑话的笑点在哪？")
fmt.Println("-----------------")
// Create the second-round conversation request
second_resp, second_err := client.CreateResponses(ctx, &responses.ResponsesRequest{
Model: "doubao-seed-2-1-pro-260915",
PreviousResponseID: responses.NewOptString(id),
Input: responses.NewInputItemArrayResponsesInput([]responses.InputItem{
inputMessage,
}),
})
if second_err != nil {
fmt.Printf("second response error: %v\n", second_err)
return
}
fmt.Println(second_resp)
}
func easyMessage(role responses.MessageRole, content string) responses.InputItem {
return responses.InputItem{
OneOf: responses.NewItemEasyMessageInputItemSum(
responses.ItemEasyMessage{
Role: responses.NewOptMessageRole(role),
Content: responses.NewStringMessageContent(content),
},
),
}
}
package com.ark.example;
import com.volcengine.ark.runtime.service.ArkService;
import com.volcengine.ark.runtime.models.responses.*;
public class demo {
public static void main(String[] args) {
String apiKey = System.getenv("ARK_API_KEY");
// The base URL for model invocation
ArkService arkService = ArkService.builder().apiKey(apiKey)
.baseUrl("https://ark.cn-beijing.volces.com/api/v3")
.build();
// Create the first-round conversation request
ResponsesRequest request = ResponsesRequest.builder()
.model("doubao-seed-2-1-pro-260915")
.input(ResponsesInput.ofString("Hi，帮我讲个笑话。"))
.build();
Response resp = arkService.createResponse(request);
System.out.println(resp);
// Create the second-round conversation request
ResponsesRequest request2 = ResponsesRequest.builder()
.model("doubao-seed-2-1-pro-260915")
.previousResponseId(resp.getId())
.input(ResponsesInput.ofList(java.util.Arrays.asList(ItemEasyMessage.builder().role(MessageRole.USER).content(
MessageContent.ofString("这个笑话的笑点在哪？")
).build())))
.build();
Response resp2 = arkService.createResponse(request2);
System.out.println(resp2);
arkService.shutdownExecutor();
}
}
from openai import OpenAI
import os
# Get API Key: https://ark.volcengine.com/region:cn-beijing/apikey
api_key = os.getenv('ARK_API_KEY')
client = OpenAI(
base_url='https://ark.cn-beijing.volces.com/api/v3',
api_key=api_key,
)
# Create the first-round conversation request
response = client.responses.create(
model="doubao-seed-2-1-pro-260915",
input="Hi，帮我讲个笑话。"
)
print(response)
# Create the second-round conversation request
second_response = client.responses.create(
model="doubao-seed-2-1-pro-260915",
previous_response_id=response.id,
input=[{"role": "user", "content": "这个笑话的笑点在哪？"}],
)
print(second_response)
上下文长度控制
输入长内容，或对话轮数增加，需同时考虑模型输出长度和上下文窗口限制。模型输入和输出会进行计量，达到长度限制会截断或报错。合理控制模型输出长度，平衡业务效果、成本与稳定性：
- 减少触发限流（TPM 限制、突发流量限制等），保障服务稳定性。
- 精确控制 token 用量，平衡成本与质量。
- 控制推理耗时，提升用户交互体验。
同时，平台支持对模型输出（思维链、回答）长度控制，来控制 token 用量。核心规格及参数，如下图：
控制输出（回答+思维链）长度
doubao-seed-translation-250915 模型不支持该字段。
Curl
Python
Go
Java
curl https://ark.cn-beijing.volces.com/api/v3/chat/completions \
-H "Content-Type: application/json" \
-H "Authorization: Bearer $ARK_API_KEY" \
-d '{
"model": "doubao-seed-2-1-pro-260915",
"messages": [
{
"role": "user",
"content": "你好"
}
],
"max_completion_tokens": 1024
}'
- 您可按需替换 Model ID。Model ID 查询见 模型列表。
import os
# Install SDK: python -m pip install --upgrade arkruntime
from arkruntime import Ark
client = Ark(
# The base URL for model invocation
base_url="https://ark.cn-beijing.volces.com/api/v3",
# Get API Key: https://ark.volcengine.com/region:cn-beijing/apikey
api_key=os.getenv('ARK_API_KEY'),
# Deep thinking takes longer; set a larger timeout, with 1,800 seconds or more recommended
timeout=1800,
)
# 创建一个对话请求
completion = client.chat.completions.create(
# Replace with Model ID
model = "doubao-seed-2-1-pro-260915",
messages=[
{"role": "system", "content": "你是 AI 人工智能助手"},
{"role": "user", "content": "常见的十字花科植物有哪些？"},
],
# 设置模型最大输出长度为 1024 token，按需调整
max_completion_tokens = 1024,
)
print(completion.choices[0].message.content)
package main
import (
"time"
"context"
"fmt"
"os"
"github.com/volcengine/ark-runtime-go/arkruntime"
"github.com/volcengine/ark-runtime-go/arkruntime/model/chat"
)
func main() {
client := arkruntime.NewClientWithApiKey(
os.Getenv("ARK_API_KEY"),
// The base URL for model invocation
arkruntime.WithBaseUrl("https://ark.cn-beijing.volces.com/api/v3"),
//深度思考耗时更长，请设置更大的超时限制，推荐为30分钟及以上
arkruntime.WithTimeout(30*time.Minute),
)
ctx := context.Background()
fmt.Println("----- standard request -----")
req := &chat.ChatCompletionRequest{
// Replace with Model ID
Model: "doubao-seed-2-1-pro-260915",
Messages: []chat.ChatCompletionRequestMessage{
systemMessage("你是 AI 人工智能助手"),
userMessage("常见的十字花科植物有哪些？"),
},
MaxCompletionTokens: chat.NewOptInt32(int32(1024)), // 设置最大输出长度为 1024 token
}
resp, err := client.CreateChatCompletion(ctx, req)
if err != nil {
fmt.Printf("standard chat error: %v\n", err)
return
}
fmt.Println(resp.Choices[0].Message.Content)
}
func systemMessage(content string) chat.ChatCompletionRequestMessage {
return chat.ChatCompletionRequestMessage{
OneOf: chat.NewChatCompletionRequestSystemMessageChatCompletionRequestMessageSum(
chat.ChatCompletionRequestSystemMessage{
Role: chat.ChatCompletionRequestSystemMessageRoleSystem,
Content: chat.NewStringChatCompletionMessageContent(content),
},
),
}
}
func userMessage(content string) chat.ChatCompletionRequestMessage {
return chat.ChatCompletionRequestMessage{
OneOf: chat.NewChatCompletionRequestUserMessageChatCompletionRequestMessageSum(
chat.ChatCompletionRequestUserMessage{
Role: chat.ChatCompletionRequestUserMessageRoleUser,
Content: chat.NewOptNilChatCompletionMessageContent(chat.NewStringChatCompletionMessageContent(content)),
},
),
}
}
package com.volcengine.ark.runtime;
import com.volcengine.ark.runtime.models.chat.*;
import com.volcengine.ark.runtime.service.ArkService;
import java.util.ArrayList;
import java.util.List;
import java.time.Duration;
public class ChatCompletionsExample {
public static void main(String[] args) {
String apiKey = System.getenv("ARK_API_KEY");
// 创建ArkService实例
ArkService arkService = ArkService.builder()
.apiKey(apiKey)
.timeout(Duration.ofMinutes(30))// 深度思考耗时更长，请设置更大的超时限制，推荐为30分钟及以上
// The base URL for model invocation
.baseUrl("https://ark.cn-beijing.volces.com/api/v3")
.build();
System.out.println("\n----- standard request -----");
final List<ChatCompletionRequestMessage> messages = new ArrayList<>();
final ChatCompletionRequestMessage systemMessage = ChatCompletionRequestSystemMessage.builder().role(ChatCompletionRequestMessageType.SYSTEM).content(ChatCompletionMessageContent.ofString("你是 AI 人工智能助手")).build();
final ChatCompletionRequestMessage userMessage = ChatCompletionRequestUserMessage.builder().role(ChatCompletionRequestMessageType.USER).content(ChatCompletionMessageContent.ofString("常见的十字花科植物有哪些？")).build();
messages.add(systemMessage);
messages.add(userMessage);
ChatCompletionRequest chatCompletionRequest = ChatCompletionRequest.builder()
.model("doubao-seed-2-1-pro-260915")//Replace with Model ID
.messages(messages)
.maxCompletionTokens(1024)// 设置最大输出长度为 1024 token
.build();
arkService.createChatCompletion(chatCompletionRequest).getChoices().forEach(choice -> System.out.println(choice.getMessage().getContent()));
// shutdown service
arkService.shutdownExecutor();
}
}
说明
- max_tokens、max_completion_tokens 不可同时设置，会直接报错。
控制回答长度
注意
DeepSeek-V4.1-flash 模型max_tokens：控制模型回答和模型思维链内容长度，超出长度后不报错，默认值 128k。
Curl
Python
Go
Java
curl https://ark.cn-beijing.volces.com/api/v3/chat/completions \
-H "Content-Type: application/json" \
-H "Authorization: Bearer $ARK_API_KEY" \
-d '{
"model": "doubao-seed-2-1-pro-260915",
"messages": [
{
"role": "user",
"content": "你好"
}
],
"max_tokens": 1024
}'
- 按需替换 Model ID，查询 Model ID 请参见 模型列表。
import os
# Install SDK: python -m pip install --upgrade arkruntime
from arkruntime import Ark
# 初始化Ark客户端
client = Ark(
# The base URL for model invocation
base_url="https://ark.cn-beijing.volces.com/api/v3",
# Get API Key: https://ark.volcengine.com/region:cn-beijing/apikey
api_key=os.getenv('ARK_API_KEY'),
)
completion = client.chat.completions.create(
# Replace with Model ID
model = "doubao-seed-2-1-pro-260915",
messages=[
{"role": "system", "content": "你是 AI 人工智能助手"},
{"role": "user", "content": "常见的十字花科植物有哪些？"},
],
# 设置模型最大输出长度为 1024 token，您可按需进行调整
max_tokens=1024,
)
print(completion.choices[0].message.content)
package main
import (
"context"
"fmt"
"github.com/volcengine/ark-runtime-go/arkruntime"
"github.com/volcengine/ark-runtime-go/arkruntime/model/chat"
"os"
)
func main() {
client := arkruntime.NewClientWithApiKey(
os.Getenv("ARK_API_KEY"),
// The base URL for model invocation
arkruntime.WithBaseUrl("https://ark.cn-beijing.volces.com/api/v3"),
)
ctx := context.Background()
fmt.Println("----- standard request -----")
req := &chat.ChatCompletionRequest{
// Replace with Model ID
Model: "doubao-seed-2-1-pro-260915",
Messages: []chat.ChatCompletionRequestMessage{
systemMessage("你是 AI 人工智能助手"),
userMessage("常见的十字花科植物有哪些？"),
},
MaxTokens: chat.NewOptInt32(1024),
}
resp, err := client.CreateChatCompletion(ctx, req)
if err != nil {
fmt.Printf("standard chat error: %v\n", err)
return
}
fmt.Println(resp.Choices[0].Message.Content)
}
func systemMessage(content string) chat.ChatCompletionRequestMessage {
return chat.ChatCompletionRequestMessage{
OneOf: chat.NewChatCompletionRequestSystemMessageChatCompletionRequestMessageSum(
chat.ChatCompletionRequestSystemMessage{
Role: chat.ChatCompletionRequestSystemMessageRoleSystem,
Content: chat.NewStringChatCompletionMessageContent(content),
},
),
}
}
func userMessage(content string) chat.ChatCompletionRequestMessage {
return chat.ChatCompletionRequestMessage{
OneOf: chat.NewChatCompletionRequestUserMessageChatCompletionRequestMessageSum(
chat.ChatCompletionRequestUserMessage{
Role: chat.ChatCompletionRequestUserMessageRoleUser,
Content: chat.NewOptNilChatCompletionMessageContent(chat.NewStringChatCompletionMessageContent(content)),
},
),
}
}
package com.ark.sample;
import com.volcengine.ark.runtime.models.chat.*;
import com.volcengine.ark.runtime.service.ArkService;
import java.util.ArrayList;
import java.util.List;
public class ChatCompletionsExample {
public static void main(String[] args) {
String apiKey = System.getenv("ARK_API_KEY");
// The base URL for model invocation
ArkService service = ArkService.builder().apiKey(apiKey)
.baseUrl("https://ark.cn-beijing.volces.com/api/v3")
.build();
System.out.println("----- standard request -----");
final List<ChatCompletionRequestMessage> messages = new ArrayList<>();
final ChatCompletionRequestMessage systemMessage = ChatCompletionRequestSystemMessage.builder().role(ChatCompletionRequestMessageType.SYSTEM).content(ChatCompletionMessageContent.ofString("你是 AI 人工智能助手")).build();
final ChatCompletionRequestMessage userMessage = ChatCompletionRequestUserMessage.builder().role(ChatCompletionRequestMessageType.USER).content(ChatCompletionMessageContent.ofString("常见的十字花科植物有哪些？")).build();
messages.add(systemMessage);
messages.add(userMessage);
ChatCompletionRequest chatCompletionRequest = ChatCompletionRequest.builder()
.model("doubao-seed-2-1-pro-260915")//Replace with Model ID
.messages(messages)
.maxTokens(1024)// 设置最大输出长度为 1024 token
.build();
service.createChatCompletion(chatCompletionRequest).getChoices().forEach(choice -> System.out.println(choice.getMessage().getContent()));
// shutdown service
service.shutdownExecutor();
}
}
控制思维链长度
设置 reasoning_effort 调节深度思考的程度，间接控制思维链内容长度。当前提供 7 档：
- none：关闭思考。
- minimal：关闭思考，直接回答。
- low：轻量思考，侧重快速响应。
- medium：均衡模式，兼顾速度与深度。
- high：深度分析，处理复杂问题。
- xhigh：更高强度思考。
- max：最高强度思考，适配高难度推理任务。
与 thinking.type 关系：
- thinking.type取值为enabled：支持配置reasoning_effort。当reasoning_effort取值为minimal时，则关闭思考，直接回答。
- thinking.type取值为disabled：reasoning_effort仅支持取值minimal。当reasoning_effort取值为其他取值时会报错。
Curl
Python
Go
Java
OpenAI SDK
curl https://ark.cn-beijing.volces.com/api/v3/chat/completions \
-H "Content-Type: application/json" \
-H "Authorization: Bearer $ARK_API_KEY" \
-d '{
"model": "doubao-seed-2-1-pro-260915",
"messages": [
{
"role": "user",
"content": [
{
"image_url": {
"url": "https://ark-project.tos-cn-beijing.ivolces.com/images/view.jpeg"
},
"type": "image_url"
},
{
"text": "图片主要讲了什么?",
"type": "text"
}
]
}
],
"thinking":{"type":"enabled"},
"reasoning_effort": "low"
}'
import os
# Install SDK: python -m pip install --upgrade arkruntime
from arkruntime import Ark
client = Ark(
# The base URL for model invocation
base_url="https://ark.cn-beijing.volces.com/api/v3",
# Get API Key: https://ark.volcengine.com/region:cn-beijing/apikey
api_key=os.getenv('ARK_API_KEY'),
# Deep thinking takes longer; set a larger timeout, with 1,800 seconds or more recommended
timeout=1800,
)
completion = client.chat.completions.create(
# Replace with Model ID .
model = "doubao-seed-2-1-pro-260915",
messages=[
{
"role": "user",
"content": [
{"type": "image_url","image_url": {"url": "https://ark-project.tos-cn-beijing.ivolces.com/images/view.jpeg"}},
{"type": "text", "text": "图片主要讲了什么?"},
],
}
],
thinking={"type":"enabled"},
reasoning_effort="low"
)
print(completion.choices[0])
package main
import (
"context"
"fmt"
"os"
"github.com/volcengine/ark-runtime-go/arkruntime"
"github.com/volcengine/ark-runtime-go/arkruntime/model/chat"
)
func main() {
client := arkruntime.NewClientWithApiKey(
//通过 os.Getenv 从环境变量中获取 ARK_API_KEY
os.Getenv("ARK_API_KEY"),
// The base URL for model invocation .
arkruntime.WithBaseUrl("https://ark.cn-beijing.volces.com/api/v3"),
)
// 创建一个上下文，通常用于传递请求的上下文信息，如超时、取消等
ctx := context.Background()
contentParts := []chat.ChatCompletionContentPart{
chat.ChatCompletionContentPart{OneOf: chat.NewChatCompletionContentPartImageChatCompletionContentPartSum(chat.ChatCompletionContentPartImage{Type: chat.ChatCompletionContentPartImageTypeImageURL, ImageURL: chat.NewChatCompletionContentPartImageImageUrlChatCompletionContentPartImageImageURL(chat.ChatCompletionContentPartImageImageUrl{URL: chat.NewOptString("https://ark-project.tos-cn-beijing.volces.com/images/view.jpeg")})})},
chat.ChatCompletionContentPart{OneOf: chat.NewChatCompletionContentPartTextChatCompletionContentPartSum(chat.ChatCompletionContentPartText{Type: chat.ChatCompletionContentPartTextTypeText, Text: "图片主要讲了什么?"})},
}
effort := chat.ReasoningEffortLow
req := &chat.ChatCompletionRequest{
// Replace with Model ID
Model: "doubao-seed-2-1-pro-260915",
Messages: []chat.ChatCompletionRequestMessage{
chat.ChatCompletionRequestMessage{OneOf: chat.NewChatCompletionRequestUserMessageChatCompletionRequestMessageSum(chat.ChatCompletionRequestUserMessage{Role: chat.ChatCompletionRequestUserMessageRoleUser, Content: chat.NewOptNilChatCompletionMessageContent(chat.NewChatCompletionContentPartArrayChatCompletionMessageContent(contentParts))})},
},
Thinking: chat.NewOptThinking(chat.Thinking{Type: chat.NewOptThinkingMode(chat.ThinkingModeEnabled)}),
ReasoningEffort: chat.NewOptReasoningEffort(effort),
}
// 发送聊天完成请求，并将结果存储在 resp 中，将可能出现的错误存储在 err 中
resp, err := client.CreateChatCompletion(ctx, req)
if err != nil {
// 若出现错误，打印错误信息并终止程序
fmt.Printf("standard chat error: %v\n", err)
return
}
// 打印聊天完成请求的响应结果
fmt.Println(resp.Choices[0].Message.Content)
}
package com.ark.sample;
import com.volcengine.ark.runtime.models.chat.*;
import com.volcengine.ark.runtime.models.chat.ChatCompletionContentPart.*;
import com.volcengine.ark.runtime.service.ArkService;
import java.util.ArrayList;
import java.util.List;
import java.util.concurrent.TimeUnit;
import okhttp3.ConnectionPool;
import okhttp3.Dispatcher;
public class MultiImageSample {
static String apiKey = System.getenv("ARK_API_KEY");
static ConnectionPool connectionPool = new ConnectionPool(5, 1, TimeUnit.SECONDS);
static Dispatcher dispatcher = new Dispatcher();
static ArkService arkService = ArkService.builder()
.dispatcher(dispatcher)
.connectionPool(connectionPool)
.baseUrl("https://ark.cn-beijing.volces.com/api/v3")
// The base URL for model invocation .
.apiKey(apiKey)
.build();
public static void main(String[] args) throws Exception {
List<ChatCompletionRequestMessage> messagesForReqList = new ArrayList<>();
// 构建消息内容
List<ChatCompletionContentPart> contentParts = new ArrayList<>();
// 第一张图片部分使用builder模式
contentParts.add(ChatCompletionContentPartImage.builder()
.type(ChatCompletionContentPartType.IMAGE_URL)
.imageUrl(ChatCompletionContentPartImageImageUrl2.ofChatCompletionContentPartImageImageUrl(ChatCompletionContentPartImageImageUrl.builder().url("https://ark-project.tos-cn-beijing.volces.com/images/view.jpeg").build()))
.build());
contentParts.add(ChatCompletionContentPartText.builder()
.type(ChatCompletionContentPartType.TEXT)
.text("图片主要讲了什么？")
.build());
messagesForReqList.add(ChatCompletionRequestUserMessage.builder()
.role(ChatCompletionRequestMessageType.USER)
.content(ChatCompletionMessageContent.ofList(contentParts))
.build());
ChatCompletionRequest req = ChatCompletionRequest.builder()
.model("doubao-seed-2-1-pro-260915") //Replace with Model ID .
.messages(messagesForReqList)
.thinking(Thinking.builder().type(ThinkingMode.ENABLED).build())
.reasoningEffort(ReasoningEffort.LOW)
.build();
arkService.createChatCompletion(req)
.getChoices()
.forEach(choice -> System.out.println(choice.getMessage().getContent()));
// shutdown service after all requests are finished
arkService.shutdownExecutor();
}
}
import os
from openai import OpenAI
client = OpenAI(
# The base URL for model invocation .
base_url="https://ark.cn-beijing.volces.com/api/v3",
# Get API Key: https://ark.volcengine.com/region:cn-beijing/apikey
api_key=os.getenv('ARK_API_KEY'),
)
completion = client.chat.completions.create(
# Replace with Model ID .
model = "doubao-seed-2-1-pro-260915",
messages=[
{
"role": "user",
"content": [
{"type": "image_url","image_url": {"url": "https://ark-project.tos-cn-beijing.ivolces.com/images/view.jpeg"}},
{"type": "text", "text": "图片主要讲了什么?"},
],
}
],
extra_body={"thinking":{"type":"enabled"}},
reasoning_effort="low"
)
print(completion.choices[0])
上下文传递逻辑
多轮对话场景
流程图 | 说明 |
流程图 | 说明 |
 |
思维链内容展现的是模型处理问题的过程，包括将问题拆分为多个问题进行处理，生成多种回复综合得出更好回答等过程。 |
工具调用场景
针对工具调用场景优化了深度思考的执行逻辑，平台会自主判断是否将思维链内容输入给模型，参与后续轮次的推理，保障结果输出连贯、准确、可解释。
流程图 | 说明 |
流程图 | 说明 |
 |
|
说明
推荐在 Responses API 中使用 previous_response_id，平台自动保存历史对话的上下文，并在多轮交互中回传给推理服务。
请求示例如下：
Chat API
Responses API
第一轮请求：触发工具调用
curl https://ark.cn-beijing.volces.com/api/v3/chat/completions \
-H "Content-Type: application/json" \
-H "Authorization: Bearer $ARK_API_KEY" \
-d '{
"model": "doubao-seed-2-0-lite-260215",
"messages": [
{
"role": "system",
"content": "你是人工智能助手。"
},
{
"role": "user",
"content": "今天北京天气怎么样"
}
],
"tools": [
{
"type": "function",
"function": {
"name": "get_weather",
"description": "天气查询",
"parameters": {
"properties": {
"location": {
"description": "地点的位置信息，例如北京、上海。",
"type": "string"
}
},
"required": [
"location"
],
"type": "object"
}
}
}
]
}'
第一轮响应：返回工具调用指令
模型会返回reasoning_content、tool_calls等关键字段。
{
"choices": [
{
"finish_reason": "tool_calls",
"index": 0,
"logprobs": null,
"message": {
"content": "",
"reasoning_content": "用户问今天北京天气怎么样，我需要调用get_weather工具，参数location填北京。检查一下，参数是必填的location，已经有了，所以直接调用工具就行。",
"role": "assistant",
"tool_calls": [
{
"function": {
"arguments": " {\"location\": \"北京\"}",
"name": "get_weather"
},
"id": "call_wiezxeyae8jzxl3jx8nhfgb5",
"type": "function"
}
]
}
}
],
...
}
第二轮请求：回传完整上下文并生成最终响应
第二轮请求还需要回传思维链信息、工具调用结果，模型生成回答。
curl https://ark.cn-beijing.volces.com/api/v3/chat/completions \
-H "Content-Type: application/json" \
-H "Authorization: Bearer $ARK_API_KEY" \
-d '{
"model": "doubao-seed-2-0-lite-260215",
"messages": [
{
"role": "system",
"content": "你是人工智能助手。"
},
{
"role": "user",
"content": "今天北京天气怎么样"
},
{
"role": "assistant",
"reasoning_content": "用户问今天北京天气怎么样，我需要调用get_weather工具，参数location填北京。检查一下，参数是必填的location，已经有了，所以直接调用工具就行。",
"tool_calls": [
{
"function": {
"arguments": " {\"location\": \"北京\"}",
"name": "get_weather"
},
"id": "call_wiezxeyae8jzxl3jx8nhfgb5",
"type": "function"
}
]
},
{
"role": "tool",
"tool_call_id":"call_wiezxeyae8jzxl3jx8nhfgb5",
"content": "5度"
}
],
"tools": [
{
"type": "function",
"function": {
"name": "get_weather",
"description": "天气查询",
"parameters": {
"properties": {
"location": {
"description": "地点的位置信息，例如北京、上海。",
"type": "string"
}
},
"required": [
"location"
],
"type": "object"
}
}
}
]
}'
第一轮请求：触发工具调用
curl https://ark.cn-beijing.volces.com/api/v3/responses \
-H "Authorization: Bearer $ARK_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "doubao-seed-2-0-lite-260215",
"input": [
{
"role": "system",
"content": "你是人工智能助手."
},
{
"role": "user",
"content": "今天北京天气怎么样"
}
],
"thinking":{"type": "enabled"},
"tools": [
{
"type": "function",
"name": "get_weather",
"description": "天气查询",
"parameters": {
"type": "object",
"properties": {
"location": {
"type": "string",
"description": "地点的位置信息，例如北京、上海。"
}
},
"required": ["location"]
}
}
]
}'
第一轮响应：返回工具调用指令
模型返回信息包含id、call_id、arguments等关键字段。
{
"created_at": 1766126702,
"id": "resp_0217661267019147d8950efa0e2f7c9d9cc7a1cc971272cf4548c",
"max_output_tokens": 32768,
"model": "doubao-seed-2-0-lite-260215",
"object": "response",
"output": [
{
"id": "rs_02176612670248500000000000000000000ffffac154e10754f5c",
"type": "reasoning",
"summary": [
{
"type": "summary_text",
"text": "用户问今天北京的天气怎么样，我需要调用get_weather工具，参数location是北京。按照格式要求来写函数调用。"
}
],
"status": "completed"
},
{
"arguments": " {\"location\": \"北京\"}",
"call_id": "call_t885uulopdd499rn0pioze7l",
"name": "get_weather",
"type": "function_call",
"id": "fc_02176612670345400000000000000000000ffffac154e10a6753e",
"status": "completed"
}
],
....
}
第二轮请求：回传结果并生成最终响应
传入上一轮 response_id、工具调用结果等信息，模型生成自然语言回答。
curl https://ark.cn-beijing.volces.com/api/v3/responses \
-H "Authorization: Bearer $ARK_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "doubao-seed-2-0-lite-260215",
"input": [
{
"type": "function_call_output",
"call_id": "call_t885uulopdd499rn0pioze7l",
"output": "5度"
}
],
"previous_response_id": "resp_0217661267019147d8950efa0e2f7c9d9cc7a1cc971272cf4548c",
"thinking":{"type": "enabled"},
"tools": [
{
"type": "function",
"name": "get_weather",
"description": "天气查询",
"parameters": {
"type": "object",
"properties": {
"location": {
"type": "string",
"description": "地点的位置信息，例如北京、上海。"
}
},
"required": ["location"]
}
}
]
}'
长度限制生效逻辑
模型输出的组成如下公式，模型的输出 = 模型的回答 + 模型的思维链（如有）
- max_tokens：控制模型的回答长度，默认值 4096。如果触发长度限制，模型停止回答，并在返回结构体的 finish_reason 字段为 length。
- max_completion_tokens：控制模型的回答、模型思维链总长度，如触发限制，模型停止回答，并在返回结构体的 finish_reason 字段为 length。如配置了 max_completion_tokens ，max_tokens默认值会失效，常用于超长文本输出或者控制模型返回长度。
仅配置 max_tokens（默认逻辑）
下面是各个对象的长度限制说明。
对象 | 关键参数 | 说明 | 可否配置 |
对象 | 关键参数 | 说明 | 可否配置 |
问题 | 最大输入长度 | 上下文窗口 - 思维链窗口，与回答共享配额。 可理解为问答配额。 | 模型规格，不可配置 |
思维链 | 思维链窗口 | 独占，不共享，即单次请求如有未使用的配额也不共享给问题及回答。 | 模型规格，不可配置 |
回答 | 最大回答长度 | 最大回答长度，通过 max_tokens 参数配置。 另外随着输入长度增加，回答配额也会受 最大输入长度（问答配额） 剩余配额影响。 回答配额 =min(最大回答长度, 最大输入长度-实际输入长度) =min(最大回答长度, 上下文窗口-思维链窗口-实际输入长度) | API字段，可配置 |
| 最大输入长度 | | 模型规格，不可配置 |
内容截断逻辑举例：模型A属性如下，上下文窗口 96k，思维链窗口 32k，最大输入长度(问答配额) 64k，配置最大回答长度 16k。
- 当用户输入的问题 56k，模型输出思维链16k，模型输出的回答达到8k时，问题+回答 = 56k+8k = 64k，触发最大输入长度（问答配额） 限制，模型停止推理。
- 当用户输入的问题 22k，模型输出思维链16k，模型输出的回答达到16k时，触发最大回答长度限制，模型停止推理。
- 当用户输入的问题 22k，模型输出思维链32k，触发最大思维链长度限制，模型停止推理。
仅配置 max_completion_tokens
影响/控制对象 | 关键参数 | 说明 | 可否配置 |
影响/控制对象 | 关键参数 | 说明 | 可否配置 |
问题 | 最大输入长度 | 与回答共享配额，即未使用配额共享给回答内容 | 模型规格，不可配置 |
思维链 | 最大输出长度 | 与回答共享配额，即未使用的配额共享给回答。 | API字段，可配置 |
| 思维链窗口 | 模型能够输出的最大思维链长度。 | 模型规格，不可配置 |
回答 | 最大输出长度 | 与思维链共享配额，即输出长度不可超过最大输出长度-已输出的思维链内容长度 | API字段，可配置 |
内容截断逻辑举例：模型A属性如下，上下文窗口 96k，最大输入长度 64k，最大思维链长度 32k。配置最大输出长度（max_completion_tokens） 32k 后，最大回答长度（max_tokens）默认值(4096)限制失效：
- 当用户输入的内容长度 72k，触发最大输入长度限制，直接报错。
- 当模型输出的内容超过 32k，触发 最大思维链长度 限制，模型停止推理。
- 当用户输入的问题长度 26k，模型输出的思维链长度 16k，模型输出的回答长度达到 16k 时，思维链长度+回答长度=16k+16k=32k，触发最大输出长度 限制，模型停止推理。
最近更新时间：2026.10.09 14:36:34
这个页面对您有帮助吗？
有用
有用
无用
无用