Skip to content

阿里通义千问API完全指南

通义千问是阿里云推出的大语言模型,与阿里云生态深度集成,适合企业级应用

概述

通义千问(Qwen)是阿里云推出的大语言模型系列,以其强大的中文能力、多模态支持和阿里云生态集成著称。本教程将带你全面掌握通义千问API的使用方法。

为什么选择通义千问?

优势说明
中文能力强专为中文优化,理解准确
多模态支持支持文本、图像、音频、视频
阿里云集成与阿里云产品深度集成
企业级服务完善的企业级支持

通义千问模型概览

通义千问模型家族:

Qwen2.5系列(最新)
├── qwen-max              # 最强能力
├── qwen-plus             # 平衡版本
├── qwen-turbo            # 快速版本
├── qwen-long             # 长上下文版本
└── qwen-vl-max           # 多模态版本

Qwen2系列(经典)
├── qwen2-72b-instruct    # 开源大模型
└── qwen2-7b-instruct     # 开源小模型

专业模型
├── qwen-coder-turbo      # 代码专用
├── qwen-math-turbo       # 数学专用
└── qwen-vl-ocr           # OCR专用

基本概念

API Key

php
<?php
// 在阿里云控制台获取API Key
// https://dashscope.console.aliyun.com/

// API Key格式
// sk-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

$apiKey = getenv('DASHSCOPE_API_KEY');

DashScope API

通义千问通过DashScope API提供服务:

php
<?php
// DashScope API地址
$baseUrl = 'https://dashscope.aliyuncs.com/compatible-mode/v1';

环境准备

创建通义千问客户端

php
<?php
require 'vendor/autoload.php';

use GuzzleHttp\Client;
use GuzzleHttp\Exception\RequestException;

class QwenClient
{
    private $client;
    private $apiKey;
    private $baseUrl = 'https://dashscope.aliyuncs.com/compatible-mode/v1';

    public function __construct(string $apiKey)
    {
        $this->apiKey = $apiKey;
        $this->client = new Client([
            'base_uri' => $this->baseUrl,
            'timeout' => 120,
            'headers' => [
                'Authorization' => 'Bearer ' . $this->apiKey,
                'Content-Type' => 'application/json',
            ],
        ]);
    }

    public function chat(
        array $messages,
        string $model = 'qwen-turbo',
        array $options = []
    ): array {
        $params = [
            'model' => $model,
            'messages' => $messages,
        ];

        if (isset($options['temperature'])) {
            $params['temperature'] = $options['temperature'];
        }

        if (isset($options['max_tokens'])) {
            $params['max_tokens'] = $options['max_tokens'];
        }

        if (isset($options['top_p'])) {
            $params['top_p'] = $options['top_p'];
        }

        try {
            $response = $this->client->post('/chat/completions', [
                'json' => $params,
            ]);

            return json_decode($response->getBody(), true);
        } catch (RequestException $e) {
            $errorBody = $e->getResponse() ? $e->getResponse()->getBody()->getContents() : 'Unknown error';
            throw new Exception('Qwen API Error: ' . $errorBody);
        }
    }
}

// 使用示例
$apiKey = getenv('DASHSCOPE_API_KEY');
$client = new QwenClient($apiKey);

$result = $client->chat([
    ['role' => 'user', 'content' => '请用一句话介绍PHP语言']
]);

echo $result['choices'][0]['message']['content'];

运行结果:

PHP是一种开源的服务器端脚本语言,特别适合Web开发,可以嵌入HTML中执行。

多模态处理

图像理解

php
<?php
class QwenVLClient
{
    private $client;
    private $apiKey;
    private $baseUrl = 'https://dashscope.aliyuncs.com/api/v1';

    public function __construct(string $apiKey)
    {
        $this->apiKey = $apiKey;
        $this->client = new Client([
            'base_uri' => $this->baseUrl,
            'timeout' => 120,
            'headers' => [
                'Authorization' => 'Bearer ' . $this->apiKey,
                'Content-Type' => 'application/json',
            ],
        ]);
    }

    public function analyzeImage(string $imageUrl, string $question): string
    {
        $response = $this->client->post('/services/aigc/multimodal-generation/generation', [
            'json' => [
                'model' => 'qwen-vl-max',
                'input' => [
                    'messages' => [
                        [
                            'role' => 'user',
                            'content' => [
                                ['image' => $imageUrl],
                                ['text' => $question],
                            ],
                        ],
                    ],
                ],
            ],
        ]);

        $result = json_decode($response->getBody(), true);
        return $result['output']['choices'][0]['message']['content'][0]['text'];
    }

    public function ocr(string $imageUrl): string
    {
        return $this->analyzeImage($imageUrl, '请识别图片中的所有文字');
    }

    public function describeImage(string $imageUrl): string
    {
        return $this->analyzeImage($imageUrl, '请详细描述这张图片的内容');
    }
}

// 使用示例
$vlClient = new QwenVLClient($apiKey);
$description = $vlClient->describeImage('https://example.com/image.jpg');
echo $description;

高级参数配置

完整参数示例

php
<?php
class QwenClient
{
    // ... 前面的代码 ...

    public function chatAdvanced(
        array $messages,
        string $model = 'qwen-turbo',
        array $options = []
    ): array {
        $params = [
            'model' => $model,
            'messages' => $messages,
        ];

        $optionalParams = [
            'temperature',
            'max_tokens',
            'top_p',
            'top_k',
            'stop',
            'seed',
            'repetition_penalty',
        ];

        foreach ($optionalParams as $param) {
            if (isset($options[$param])) {
                $params[$param] = $options[$param];
            }
        }

        // 启用搜索增强
        if (!empty($options['enable_search'])) {
            $params['enable_search'] = true;
        }

        try {
            $response = $this->client->post('/chat/completions', [
                'json' => $params,
            ]);

            return json_decode($response->getBody(), true);
        } catch (RequestException $e) {
            $errorBody = $e->getResponse() ? $e->getResponse()->getBody()->getContents() : 'Unknown error';
            throw new Exception('Qwen API Error: ' . $errorBody);
        }
    }
}

// 使用示例
$result = $client->chatAdvanced(
    [['role' => 'user', 'content' => '今天有什么新闻?']],
    'qwen-turbo',
    [
        'temperature' => 0.7,
        'enable_search' => true, // 启用联网搜索
    ]
);

参数详解

参数范围默认值说明
temperature0-21控制随机性
max_tokens1-60001500最大输出Token
top_p0-10.8核采样参数
top_k0-1000只考虑前K个候选词
seed整数-随机种子
repetition_penalty1-21.1重复惩罚
enable_searchbooleanfalse启用联网搜索

流式响应处理

php
<?php
class QwenClient
{
    // ... 前面的代码 ...

    public function chatStream(
        array $messages,
        string $model = 'qwen-turbo'
    ): Generator {
        $response = $this->client->post('/chat/completions', [
            'json' => [
                'model' => $model,
                'messages' => $messages,
                'stream' => true,
            ],
            'stream' => true,
        ]);

        $body = $response->getBody();
        $buffer = '';

        while (!$body->eof()) {
            $chunk = $body->read(1024);
            $buffer .= $chunk;

            while (($pos = strpos($buffer, "\n")) !== false) {
                $line = substr($buffer, 0, $pos);
                $buffer = substr($buffer, $pos + 1);

                $line = trim($line);
                if (empty($line) || $line === 'data: [DONE]') {
                    continue;
                }

                if (strpos($line, 'data: ') === 0) {
                    $json = substr($line, 6);
                    $data = json_decode($json, true);

                    if (isset($data['choices'][0]['delta']['content'])) {
                        yield $data['choices'][0]['delta']['content'];
                    }
                }
            }
        }
    }
}

// 使用示例
echo "通义千问回复:";
foreach ($client->chatStream([['role' => 'user', 'content' => '讲一个程序员笑话']]) as $chunk) {
    echo $chunk;
    flush();
}

长文本处理

使用qwen-long处理长文档

php
<?php
class LongDocumentProcessor
{
    private QwenClient $client;

    public function __construct(QwenClient $client)
    {
        $this->client = $client;
    }

    public function analyzeDocument(string $document, string $question): string
    {
        $result = $this->client->chat(
            [
                [
                    'role' => 'user',
                    'content' => "文档内容:\n{$document}\n\n问题:{$question}"
                ]
            ],
            'qwen-long',
            ['max_tokens' => 4096]
        );

        return $result['choices'][0]['message']['content'];
    }

    public function summarizeDocument(string $document, int $maxLength = 500): string
    {
        $result = $this->client->chat(
            [
                [
                    'role' => 'user',
                    'content' => "请将以下文档总结为不超过{$maxLength}字:\n\n{$document}"
                ]
            ],
            'qwen-long'
        );

        return $result['choices'][0]['message']['content'];
    }
}

常见错误与踩坑点

错误1:模型名称错误

php
<?php
// ❌ 错误做法:使用不存在的模型
$result = $client->chat($messages, 'qwen-3');

// ✅ 正确做法:使用正确的模型名称
$result = $client->chat($messages, 'qwen-turbo');
$result = $client->chat($messages, 'qwen-plus');
$result = $client->chat($messages, 'qwen-max');

错误2:忽略API配额

php
<?php
// ❌ 错误做法:不处理配额限制
$result = $client->chat($messages);

// ✅ 正确做法:处理配额错误
try {
    $result = $client->chat($messages);
} catch (Exception $e) {
    if (strpos($e->getMessage(), 'QuotaExhausted') !== false) {
        echo 'API配额已用尽,请检查账户余额';
    }
}

错误3:忽略enable_search选项

php
<?php
// ❌ 错误做法:不启用搜索获取实时信息
$result = $client->chat([['role' => 'user', 'content' => '今天天气如何?']]);

// ✅ 正确做法:启用联网搜索
$result = $client->chatAdvanced(
    [['role' => 'user', 'content' => '今天天气如何?']],
    'qwen-turbo',
    ['enable_search' => true]
);

常见应用场景

场景1:智能客服

php
<?php
class QwenCustomerService
{
    private QwenClient $client;
    private array $knowledgeBase = [];

    public function addKnowledge(string $category, string $content): void
    {
        $this->knowledgeBase[$category] = $content;
    }

    public function handleQuery(string $query): string
    {
        $context = '';
        foreach ($this->knowledgeBase as $category => $content) {
            $context .= "【{$category}】\n{$content}\n\n";
        }

        $result = $this->client->chat([
            [
                'role' => 'system',
                'content' => '你是一个友好的客服助手。请根据知识库回答问题。'
            ],
            [
                'role' => 'user',
                'content' => "知识库:\n{$context}\n\n客户问题:{$query}"
            ]
        ]);

        return $result['choices'][0]['message']['content'];
    }
}

场景2:文档问答

php
<?php
class QwenDocumentQA
{
    private QwenClient $client;
    private string $document;

    public function loadDocument(string $filePath): void
    {
        $this->document = file_get_contents($filePath);
    }

    public function ask(string $question): string
    {
        $result = $this->client->chat(
            [
                [
                    'role' => 'user',
                    'content' => "文档内容:\n{$this->document}\n\n问题:{$question}"
                ]
            ],
            'qwen-long'
        );

        return $result['choices'][0]['message']['content'];
    }
}

场景3:代码助手

php
<?php
class QwenCodeAssistant
{
    private QwenClient $client;

    public function generateCode(string $description, string $language = 'PHP'): string
    {
        $result = $this->client->chat(
            [
                [
                    'role' => 'user',
                    'content' => "请生成{$language}代码:{$description}"
                ]
            ],
            'qwen-coder-turbo'
        );

        return $result['choices'][0]['message']['content'];
    }
}

场景4:内容创作

php
<?php
class QwenContentCreator
{
    private QwenClient $client;

    public function generateArticle(string $topic, int $wordCount = 800): string
    {
        $result = $this->client->chat([
            [
                'role' => 'user',
                'content' => "请写一篇关于{$topic}的文章,字数约{$wordCount}字。"
            ]
        ]);

        return $result['choices'][0]['message']['content'];
    }
}

场景5:图像分析

php
<?php
class QwenImageAnalyzer
{
    private QwenVLClient $client;

    public function describeImage(string $imageUrl): string
    {
        return $this->client->describeImage($imageUrl);
    }

    public function extractText(string $imageUrl): string
    {
        return $this->client->ocr($imageUrl);
    }

    public function analyzeChart(string $imageUrl): string
    {
        return $this->client->analyzeImage($imageUrl, '请分析这张图表的数据和趋势');
    }
}

常见问题答疑(FAQ)

Q1:通义千问的定价如何?

回答

模型输入价格输出价格
qwen-turbo¥0.3/百万Token¥0.6/百万Token
qwen-plus¥0.8/百万Token¥2/百万Token
qwen-max¥2/百万Token¥6/百万Token
qwen-long¥0.5/百万Token¥2/百万Token

Q2:如何选择合适的模型?

回答

场景推荐模型原因
快速响应qwen-turbo速度快,成本低
复杂任务qwen-max能力最强
长文档qwen-long支持长上下文
代码生成qwen-coder-turbo代码专用
图像理解qwen-vl-max多模态支持

Q3:如何启用联网搜索?

回答

php
<?php
$result = $client->chatAdvanced(
    [['role' => 'user', 'content' => '今天有什么新闻?']],
    'qwen-turbo',
    ['enable_search' => true]
);

Q4:如何处理API错误?

回答

php
<?php
function handleQwenError(Exception $e): string
{
    $message = $e->getMessage();

    if (strpos($message, 'InvalidApiKey') !== false) {
        return 'API Key无效';
    }
    if (strpos($message, 'QuotaExhausted') !== false) {
        return 'API配额已用尽';
    }
    if (strpos($message, 'RateLimitExceeded') !== false) {
        return '请求过于频繁';
    }

    return '服务暂时不可用';
}

Q5:如何使用阿里云SDK?

回答

php
<?php
// 安装阿里云SDK
// composer require alibabacloud/dashscope-sdk-php

use AlibabaCloud\SDK\Dashscope\Dashscope;
use AlibabaCloud\SDK\Dashscope\Models\ChatCompletionRequest;
use AlibabaCloud\Tea\Utils\Utils\RuntimeOptions;

$client = new Dashscope();
$request = new ChatCompletionRequest([
    'model' => 'qwen-turbo',
    'messages' => [
        ['role' => 'user', 'content' => '你好'],
    ],
]);

$response = $client->chatCompletion($request, new RuntimeOptions());
echo $response->body->choices[0]->message->content;

Q6:如何实现多模态输入?

回答

php
<?php
// 使用qwen-vl-max进行图像理解
$vlClient = new QwenVLClient($apiKey);
$result = $vlClient->analyzeImage('https://example.com/image.jpg', '这张图片里有什么?');

实战练习

基础练习

练习1:编写一个简单的通义千问聊天程序。

参考代码

php
<?php
$apiKey = getenv('DASHSCOPE_API_KEY');
$client = new QwenClient($apiKey);

echo "通义千问聊天助手 (输入 'quit' 退出)\n";

while (true) {
    echo "\n你: ";
    $input = trim(fgets(STDIN));

    if ($input === 'quit') {
        break;
    }

    $result = $client->chat([['role' => 'user', 'content' => $input]]);
    echo "通义千问: " . $result['choices'][0]['message']['content'] . "\n";
}

进阶练习

练习2:实现一个图像描述生成器。

参考代码

php
<?php
class ImageCaptionGenerator
{
    private QwenVLClient $client;

    public function generate(string $imageUrl, string $style = '简洁'): string
    {
        $prompt = "请用{$style}的语言描述这张图片:";
        return $this->client->analyzeImage($imageUrl, $prompt);
    }
}

挑战练习

练习3:构建一个多模态文档分析系统。

参考代码

php
<?php
class MultimodalDocumentAnalyzer
{
    private QwenClient $textClient;
    private QwenVLClient $vlClient;

    public function analyze(array $content): array
    {
        $results = [];

        foreach ($content as $item) {
            if ($item['type'] === 'text') {
                $results[] = $this->analyzeText($item['data']);
            } elseif ($item['type'] === 'image') {
                $results[] = $this->analyzeImage($item['data']);
            }
        }

        return $results;
    }

    private function analyzeText(string $text): array
    {
        $result = $this->textClient->chat([
            ['role' => 'user', 'content' => "请分析以下文本:\n{$text}"]
        ]);

        return ['type' => 'text', 'analysis' => $result['choices'][0]['message']['content']];
    }

    private function analyzeImage(string $imageUrl): array
    {
        return ['type' => 'image', 'analysis' => $this->vlClient->describeImage($imageUrl)];
    }
}

知识点总结

核心要点

  1. 中文能力强:专为中文优化,理解准确
  2. 多模态支持:支持文本、图像、音频、视频
  3. 阿里云集成:与阿里云产品深度集成
  4. 联网搜索:支持enable_search获取实时信息
  5. 长文本处理:qwen-long支持长上下文

易错点回顾

易错点正确做法
模型名称错误使用qwen-turbo/qwen-plus/qwen-max
忽略API配额处理QuotaExhausted错误
不启用搜索使用enable_search获取实时信息
不使用专用模型代码任务使用qwen-coder-turbo

拓展参考资料

官方文档

进阶学习路径

  1. 本知识点 → 通义千问API基础使用
  2. 下一步智谱AI GLM
  3. 进阶错误处理与重试
  4. 高级安全与鉴权

💡 记住:通义千问与阿里云生态深度集成,适合需要企业级服务和阿里云产品的场景。善用enable_search可以获取实时信息。