OpenSearch:Content generation LLM

Prerequisites

Before you begin, ensure that you have:

• An activated AI Search Open Platform service. See Activate the service

• An API key for authentication. See Obtain an API key. Your API key is in the format OS-**** and is passed as a Bearer token when initializing the client

Usage notes

• The request body cannot exceed 8 MB.

• For a complete list of parameters, see Content generation service.

Send a single-turn request

The following example shows the minimum code required to send one message and print the response.

package com.aliyun.sample;

import com.aliyun.searchplat20240529.Client;
import com.aliyun.searchplat20240529.models.GetTextGenerationRequest;
import com.aliyun.searchplat20240529.models.GetTextGenerationResponse;
import com.aliyun.teaopenapi.models.Config;
import com.fasterxml.jackson.databind.ObjectMapper;

import java.util.ArrayList;
import java.util.List;

public class TextGenerationDemo {
    public static void main(String[] args) throws Exception {
        Config config = new Config();
        // Pass your API key as a Bearer token
        config.setBearerToken("your-api-key: OS-****");
        // Endpoint: the unified request endpoint, without the http:// prefix
        config.setEndpoint("***.opensearch.aliyuncs.com");
        config.setProtocol("http");
        Client client = new Client(config);

        final ObjectMapper objectMapper = new ObjectMapper();
        List<GetTextGenerationRequest.GetTextGenerationRequestMessages> messages =
            objectMapper.readValue("[]", List.class);

        messages.add(
            new GetTextGenerationRequest.GetTextGenerationRequestMessages()
                .setRole("user")
                .setContent("What are some fun things to do in Shanghai?")
        );

        GetTextGenerationRequest request = new GetTextGenerationRequest();
        request.setMessages(messages);

        GetTextGenerationResponse response = client.getTextGeneration("default", "ops-qwen-turbo", request);
        System.out.println(response.getBody().getResult().getText());
    }
}

Replace the following placeholders with your actual values:

Send a multi-turn conversation

Each call to the model is stateless — it has no memory of previous exchanges. To build a conversation, pass the full message history with every request. Use the user role for the human side and the assistant role to include the model's previous responses.

package com.aliyun.sample;

import com.aliyun.searchplat20240529.Client;
import com.aliyun.searchplat20240529.models.GetTextGenerationRequest;
import com.aliyun.searchplat20240529.models.GetTextGenerationResponse;
import com.aliyun.teaopenapi.models.Config;
import com.fasterxml.jackson.databind.ObjectMapper;

import java.util.List;

public class TextGenerationDemo {
    public static void main(String[] args) throws Exception {
        Config config = new Config();
        config.setBearerToken("your-api-key: OS-****");
        config.setEndpoint("***.opensearch.aliyuncs.com");
        config.setProtocol("http");
        Client client = new Client(config);

        // Build the conversation history. Each entry has a "role" (user or assistant)
        // and "content" (the message text).
        String history = "[{\"role\":\"user\",\"content\":\"hello\"},{\"role\":\"assistant\",\"content\":\"hello\"}]";
        final ObjectMapper objectMapper = new ObjectMapper();
        List<GetTextGenerationRequest.GetTextGenerationRequestMessages> messages =
            objectMapper.readValue(history, List.class);

        // Append the next user message
        messages.add(
            new GetTextGenerationRequest.GetTextGenerationRequestMessages()
                .setRole("user")
                .setContent("What are some fun things to do in Shanghai?")
        );

        GetTextGenerationRequest request = new GetTextGenerationRequest();
        request.setMessages(messages);

        GetTextGenerationResponse response = client.getTextGeneration("default", "ops-qwen-turbo", request);
        System.out.println(response.getBody().getResult().getText());
    }
}

What's next

• Content generation service — Full parameter reference for the text generation API