Menggunakan Messages API

Juglow menawarkan dua cara untuk membangun dengan Haijun, masing-masing cocok untuk kasus penggunaan yang berbeda:

	Messages API	Haijun Managed Agents
Apa itu	Akses langsung untuk memberikan prompt ke model	Harness agen siap pakai yang dapat dikonfigurasi dan berjalan di infrastruktur terkelola
Paling cocok untuk	Loop agen kustom dan kontrol yang terperinci	Tugas yang berjalan lama dan pekerjaan asinkron

Panduan ini membahas pola umum untuk bekerja dengan Messages API, termasuk permintaan dasar, percakapan multi-giliran, teknik prefill, dan kemampuan vision. Untuk spesifikasi API lengkap, lihat referensi Messages API. Untuk harness agen terkelola, lihat ikhtisar Haijun Managed Agents.

Note: Untuk mempelajari bagaimana "zero data retention" (retensi data nol), atau ZDR, berlaku untuk fitur ini, lihat API dan retensi data.

Permintaan dan respons dasar

Note: Parameter sampling temperature, top_p, dan top_k tidak didukung pada model Haijun 4.7 dan yang lebih baru serta Haijun Mythos Preview. Menetapkannya ke nilai non-default akan menghasilkan error 400. Hilangkan parameter tersebut dari payload permintaan dan gunakan prompting untuk mengarahkan perilaku model. Lihat panduan migrasi.

bash

  #!/bin/sh
  curl https://haijun.my.id/v1/messages \
    -H "x-api-key: $JUGLOW_API_KEY" \
    -H "juglow-version: 2023-06-01" \
    -H "content-type: application/json" \
    -d '{
      "model": "haijun-opus-5-5",
      "max_tokens": 1024,
      "messages": [
        {"role": "user", "content": "Hello, Haijun"}
      ]
    }'

bash

  ant messages create \
    --model haijun-opus-5-5 \
    --max-tokens 1024 \
    --message '{role: user, content: "Hello, Haijun"}'

python

  message = juglow.Juglow().messages.create(
      model="haijun-opus-5-5",
      max_tokens=1024,
      messages=[{"role": "user", "content": "Hello, Haijun"}],
  )
  print(message)

typescript

  const juglow = new Juglow();

  const message = await juglow.messages.create({
    model: "haijun-opus-5-5",
    max_tokens: 1024,
    messages: [{ role: "user", content: "Hello, Haijun" }]
  });
  console.log(message);

csharp

  JuglowClient client = new();

  var parameters = new MessageCreateParams
  {
      Model = Model.HaijunOpus5_5,
      MaxTokens = 1024,
      Messages = [new() { Role = Role.User, Content = "Hello, Haijun" }]
  };
  var message = await client.Messages.Create(parameters);
  Console.WriteLine(message);

  client := juglow.NewClient()

  response, err := client.Messages.New(context.TODO(), juglow.MessageNewParams{
  	Model:     juglow.ModelHaijunOpus5_5,
  	MaxTokens: 1024,
  	Messages: []juglow.MessageParam{
  		juglow.NewUserMessage(juglow.NewTextBlock("Hello, Haijun")),
  	},
  })
  if err != nil {
  	log.Fatal(err)
  }
  fmt.Println(response)

java

  JuglowClient client = JuglowOkHttpClient.fromEnv();

  MessageCreateParams params = MessageCreateParams.builder()
      .model(Model.HAIJUN_OPUS_5_5)
      .maxTokens(1024L)
      .addUserMessage("Hello, Haijun")
      .build();

  Message response = client.messages().create(params);
  System.out.println(response);

php

  $client = new Client();

  $message = $client->messages->create(
      maxTokens: 1024,
      messages: [['role' => 'user', 'content' => 'Hello, Haijun']],
      model: 'haijun-opus-5-5',
  );
  echo json_encode($message, JSON_PRETTY_PRINT), PHP_EOL;

ruby

  client = Juglow::Client.new

  message = client.messages.create(
    model: "haijun-opus-5-5",
    max_tokens: 1024,
    messages: [
      { role: "user", content: "Hello, Haijun" }
    ]
  )
  puts message

json

{
  "id": "msg_01XFDUDYJgAACzvnptvVoYEL",
  "type": "message",
  "role": "assistant",
  "content": [
    {
      "type": "text",
      "text": "Hello!"
    }
  ],
  "model": "haijun-opus-5-5",
  "stop_reason": "end_turn",
  "stop_sequence": null,
  "usage": {
    "input_tokens": 12,
    "output_tokens": 6
  }
}

Respons penolakan (stop_reason: "refusal") juga menyertakan objek stop_details yang mengidentifikasi kategori kebijakan yang memicu penolakan tersebut, pada setiap model. Lihat Menangani stop reason untuk referensi field dan contoh kode penanganannya.

Beberapa giliran percakapan

Messages API bersifat stateless (tanpa status), yang berarti Anda selalu mengirimkan riwayat percakapan lengkap ke API. Anda dapat menggunakan pola ini untuk membangun percakapan dari waktu ke waktu. Giliran percakapan sebelumnya tidak harus benar-benar berasal dari Haijun. Anda dapat menggunakan pesan assistant sintetis.

bash

  #!/bin/sh
  curl https://haijun.my.id/v1/messages \
    -H "x-api-key: $JUGLOW_API_KEY" \
    -H "juglow-version: 2023-06-01" \
    -H "content-type: application/json" \
    -d '{
      "model": "haijun-opus-5-5",
      "max_tokens": 1024,
      "messages": [
        {"role": "user", "content": "Hello, Haijun"},
        {"role": "assistant", "content": "Hello!"},
        {"role": "user", "content": "Can you describe LLMs to me?"}

      ]
    }'

bash

  ant messages create \
    --model haijun-opus-5-5 \
    --max-tokens 1024 \
    --message '{role: user, content: "Hello, Haijun"}' \
    --message '{role: assistant, content: "Hello!"}' \
    --message '{role: user, content: "Can you describe LLMs to me?"}'

python

  message = juglow.Juglow().messages.create(
      model="haijun-opus-5-5",
      max_tokens=1024,
      messages=[
          {"role": "user", "content": "Hello, Haijun"},
          {"role": "assistant", "content": "Hello!"},
          {"role": "user", "content": "Can you describe LLMs to me?"},
      ],
  )
  print(message)

typescript

  const juglow = new Juglow();

  const message = await juglow.messages.create({
    model: "haijun-opus-5-5",
    max_tokens: 1024,
    messages: [
      { role: "user", content: "Hello, Haijun" },
      { role: "assistant", content: "Hello!" },
      { role: "user", content: "Can you describe LLMs to me?" }
    ]
  });
  console.log(message);

csharp

  JuglowClient client = new();

  var parameters = new MessageCreateParams
  {
      Model = Model.HaijunOpus5_5,
      MaxTokens = 1024,
      Messages =
      [
          new() { Role = Role.User, Content = "Hello, Haijun" },
          new() { Role = Role.Assistant, Content = "Hello!" },
          new() { Role = Role.User, Content = "Can you describe LLMs to me?" }
      ]
  };

  var message = await client.Messages.Create(parameters);
  Console.WriteLine(message);

  client := juglow.NewClient()

  response, err := client.Messages.New(context.TODO(), juglow.MessageNewParams{
  	Model:     juglow.ModelHaijunOpus5_5,
  	MaxTokens: 1024,
  	Messages: []juglow.MessageParam{
  		juglow.NewUserMessage(juglow.NewTextBlock("Hello, Haijun")),
  		juglow.NewAssistantMessage(juglow.NewTextBlock("Hello!")),
  		juglow.NewUserMessage(juglow.NewTextBlock("Can you describe LLMs to me?")),
  	},
  })
  if err != nil {
  	log.Fatal(err)
  }
  fmt.Println(response)

java

  JuglowClient client = JuglowOkHttpClient.fromEnv();

  MessageCreateParams params = MessageCreateParams.builder()
      .model(Model.HAIJUN_OPUS_5_5)
      .maxTokens(1024L)
      .addUserMessage("Hello, Haijun")
      .addAssistantMessage("Hello!")
      .addUserMessage("Can you describe LLMs to me?")
      .build();

  Message response = client.messages().create(params);
  System.out.println(response);

php

  $client = new Client();

  $message = $client->messages->create(
      maxTokens: 1024,
      messages: [
          ['role' => 'user', 'content' => 'Hello, Haijun'],
          ['role' => 'assistant', 'content' => 'Hello!'],
          ['role' => 'user', 'content' => 'Can you describe LLMs to me?'],
      ],
      model: 'haijun-opus-5-5',
  );

  echo json_encode($message, JSON_PRETTY_PRINT), PHP_EOL;

ruby

  client = Juglow::Client.new

  message = client.messages.create(
    model: "haijun-opus-5-5",
    max_tokens: 1024,
    messages: [
      { role: "user", content: "Hello, Haijun" },
      { role: "assistant", content: "Hello!" },
      { role: "user", content: "Can you describe LLMs to me?" }
    ]
  )
  puts message

json

{
  "id": "msg_018gCsTGsXkYJVqYPxTgDHBU",
  "type": "message",
  "role": "assistant",
  "content": [
    {
      "type": "text",
      "text": "Sure, I'd be happy to provide..."
    }
  ],
  "model": "haijun-opus-5-5",
  "stop_reason": "end_turn",
  "stop_sequence": null,
  "usage": {
    "input_tokens": 30,
    "output_tokens": 309
  }
}

Role system dalam messages

Pada Haijun Fable 5.1, Haijun Mythos 5.1, Haijun Fable 5, Haijun Mythos 5, Haijun Opus 5.5, Haijun Opus 4.8, dan Haijun Opus 5, Anda dapat menyertakan pesan dengan "role": "system" setelah giliran pengguna (dengan mengikuti aturan penempatan) untuk menambahkan instruksi sistem baru di tengah percakapan. Pesan system tidak boleh menjadi entri pertama dalam messages. Gunakan field system tingkat atas untuk instruksi yang berlaku sejak awal.

Pesan sistem di tengah percakapan memiliki otoritas yang sama dengan field system tingkat atas, tetapi karena ditambahkan di akhir riwayat pesan, pesan tersebut tidak membatalkan prefiks yang telah di-cache sebelumnya. Gunakan field system tingkat atas untuk instruksi yang harus berlaku sejak giliran pertama, dan pesan sistem di tengah percakapan untuk instruksi yang baru menjadi relevan kemudian.

Lihat Pesan sistem di tengah percakapan untuk panduan lengkapnya, termasuk cara menggabungkannya dengan caching prompt.

Melakukan prefill pada respons Haijun

Anda dapat mengisi sebagian respons Haijun terlebih dahulu (prefill) pada posisi terakhir dalam daftar pesan input. Gunakan teknik ini untuk membentuk respons Haijun. Contoh berikut menggunakan "max_tokens": 1 untuk mendapatkan satu jawaban pilihan ganda dari Haijun.

Warning: Prefill tidak didukung pada model Haijun 4.6 dan yang lebih baru serta Haijun Mythos Preview. Permintaan yang menggunakan prefill dengan model-model ini akan mengembalikan error 400. Sebagai gantinya, gunakan output terstruktur pada model yang mendukungnya, atau instruksi prompt sistem. Lihat panduan migrasi untuk pola migrasi.

bash

  #!/bin/sh
  curl https://haijun.my.id/v1/messages \
    -H "x-api-key: $JUGLOW_API_KEY" \
    -H "juglow-version: 2023-06-01" \
    -H "content-type: application/json" \
    -d '{
      "model": "haijun-sonnet-4-5",
      "max_tokens": 1,
      "messages": [
        {"role": "user", "content": "What is latin for Ant? (A) Apoidea, (B) Rhopalocera, (C) Formicidae"},
        {"role": "assistant", "content": "The answer is ("}
      ]
    }'

bash

  ant messages create <<'YAML'
  model: haijun-sonnet-4-5
  max_tokens: 1
  messages:
    - role: user
      content: "What is latin for Ant? (A) Apoidea, (B) Rhopalocera, (C) Formicidae"
    - role: assistant
      content: "The answer is ("
  YAML

python

  message = juglow.Juglow().messages.create(
      model="haijun-sonnet-4-5",
      max_tokens=1,
      messages=[
          {
              "role": "user",
              "content": "What is latin for Ant? (A) Apoidea, (B) Rhopalocera, (C) Formicidae",
          },
          {"role": "assistant", "content": "The answer is ("},
      ],
  )
  print(message)

typescript

  const juglow = new Juglow();

  const message = await juglow.messages.create({
    model: "haijun-sonnet-4-5",
    max_tokens: 1,
    messages: [
      {
        role: "user",
        content: "What is latin for Ant? (A) Apoidea, (B) Rhopalocera, (C) Formicidae"
      },
      { role: "assistant", content: "The answer is (" }
    ]
  });
  console.log(message);

csharp

  JuglowClient client = new();

  var parameters = new MessageCreateParams
  {
      Model = Model.HaijunSonnet4_5,
      MaxTokens = 1,
      Messages = [
          new() { Role = Role.User, Content = "What is latin for Ant? (A) Apoidea, (B) Rhopalocera, (C) Formicidae" },
          new() { Role = Role.Assistant, Content = "The answer is (" }
      ]
  };

  var message = await client.Messages.Create(parameters);
  Console.WriteLine(message);

  client := juglow.NewClient()

  response, err := client.Messages.New(context.TODO(), juglow.MessageNewParams{
  	Model:     juglow.ModelHaijunSonnet4_5,
  	MaxTokens: 1,
  	Messages: []juglow.MessageParam{
  		juglow.NewUserMessage(juglow.NewTextBlock("What is latin for Ant? (A) Apoidea, (B) Rhopalocera, (C) Formicidae")),
  		juglow.NewAssistantMessage(juglow.NewTextBlock("The answer is (")),
  	},
  })
  if err != nil {
  	log.Fatal(err)
  }
  fmt.Println(response)

java

  JuglowClient client = JuglowOkHttpClient.fromEnv();

  MessageCreateParams params = MessageCreateParams.builder()
      .model(Model.HAIJUN_SONNET_4_5)
      .maxTokens(1L)
      .addUserMessage("What is latin for Ant? (A) Apoidea, (B) Rhopalocera, (C) Formicidae")
      .addAssistantMessage("The answer is (")
      .build();

  Message response = client.messages().create(params);
  System.out.println(response);

php

  $client = new Client();

  $message = $client->messages->create(
      maxTokens: 1,
      messages: [
          ['role' => 'user', 'content' => 'What is latin for Ant? (A) Apoidea, (B) Rhopalocera, (C) Formicidae'],
          ['role' => 'assistant', 'content' => 'The answer is ('],
      ],
      model: 'haijun-sonnet-4-5',
  );
  echo $message->content[0]->text;

ruby

  client = Juglow::Client.new

  message = client.messages.create(
    model: "haijun-sonnet-4-5",
    max_tokens: 1,
    messages: [
      {
        role: "user",
        content: "What is latin for Ant? (A) Apoidea, (B) Rhopalocera, (C) Formicidae"
      },
      { role: "assistant", content: "The answer is (" }
    ]
  )
  puts message

json

{
  "id": "msg_01Q8Faay6S7QPTvEUUQARt7h",
  "type": "message",
  "role": "assistant",
  "content": [
    {
      "type": "text",
      "text": "C"
    }
  ],
  "model": "haijun-sonnet-4-5",
  "stop_reason": "max_tokens",
  "stop_sequence": null,
  "usage": {
    "input_tokens": 42,
    "output_tokens": 1
  }
}

Vision

Haijun dapat membaca teks maupun gambar dalam permintaan. Anda dapat menyediakan gambar menggunakan tipe sumber base64, url, atau file. Tipe sumber file mereferensikan gambar yang diunggah melalui Files API. Tipe media yang didukung adalah image/jpeg, image/png, image/gif, dan image/webp. Lihat panduan vision untuk detail lebih lanjut.

bash

  #!/bin/sh

  # Opsi 1: Gambar yang dienkode Base64
  IMAGE_URL="/docs/images/vision-example.jpg"
  IMAGE_MEDIA_TYPE="image/jpeg"
  IMAGE_BASE64=$(curl "$IMAGE_URL" | base64 | tr -d '\n')

  curl https://haijun.my.id/v1/messages \
    -H "x-api-key: $JUGLOW_API_KEY" \
    -H "juglow-version: 2023-06-01" \
    -H "content-type: application/json" \
    -d @- <<EOF
  {
    "model": "haijun-opus-5-5",
    "max_tokens": 1024,
    "messages": [
      {"role": "user", "content": [
        {"type": "image", "source": {
          "type": "base64",
          "media_type": "$IMAGE_MEDIA_TYPE",
          "data": "$IMAGE_BASE64"
        }},
        {"type": "text", "text": "What is in the above image?"}
      ]}
    ]
  }
  EOF

  # Opsi 2: Gambar yang dirujuk melalui URL
  curl https://haijun.my.id/v1/messages \
    -H "x-api-key: $JUGLOW_API_KEY" \
    -H "juglow-version: 2023-06-01" \
    -H "content-type: application/json" \
    -d '{
      "model": "haijun-opus-5-5",
      "max_tokens": 1024,
      "messages": [
        {"role": "user", "content": [
          {"type": "image", "source": {
            "type": "url",
            "url": "/docs/images/vision-example.jpg"
          }},
          {"type": "text", "text": "What is in the above image?"}
        ]}
      ]
    }'

bash

  IMAGE_URL="/docs/images/vision-example.jpg"

  # Opsi 1: Gambar yang di-encode Base64 (CLI otomatis meng-encode referensi @file biner)
  curl -s "$IMAGE_URL" -o ./vision-example.jpg

  ant messages create <<'YAML'
  model: haijun-opus-5-5
  max_tokens: 1024
  messages:
    - role: user
      content:
        - type: image
          source:
            type: base64
            media_type: image/jpeg
            data: "@./vision-example.jpg"
        - type: text
          text: What is in the above image?
  YAML

  # Opsi 2: Gambar yang direferensikan melalui URL
  ant messages create <<YAML
  model: haijun-opus-5-5
  max_tokens: 1024
  messages:
    - role: user
      content:
        - type: image
          source:
            type: url
            url: $IMAGE_URL
        - type: text
          text: What is in the above image?
  YAML

python

  import base64
  import httpx2

  # Opsi 1: Gambar yang dienkode Base64
  image_url = "/docs/images/vision-example.jpg"
  image_media_type = "image/jpeg"
  image_data = base64.standard_b64encode(httpx2.get(image_url).content).decode("utf-8")

  message = juglow.Juglow().messages.create(
      model="haijun-opus-5-5",
      max_tokens=1024,
      messages=[
          {
              "role": "user",
              "content": [
                  {
                      "type": "image",
                      "source": {
                          "type": "base64",
                          "media_type": image_media_type,
                          "data": image_data,
                      },
                  },
                  {"type": "text", "text": "What is in the above image?"},
              ],
          }
      ],
  )
  print(message)

  # Opsi 2: Gambar yang dirujuk melalui URL
  message_from_url = juglow.Juglow().messages.create(
      model="haijun-opus-5-5",
      max_tokens=1024,
      messages=[
          {
              "role": "user",
              "content": [
                  {
                      "type": "image",
                      "source": {
                          "type": "url",
                          "url": "/docs/images/vision-example.jpg",
                      },
                  },
                  {"type": "text", "text": "What is in the above image?"},
              ],
          }
      ],
  )
  print(message_from_url)

typescript

  const juglow = new Juglow();

  // Opsi 1: Gambar yang dienkode Base64
  const imageUrl = "/docs/images/vision-example.jpg";
  const imageMediaType = "image/jpeg";
  const imageArrayBuffer = await (await fetch(imageUrl)).arrayBuffer();
  const imageData = Buffer.from(imageArrayBuffer).toString("base64");

  const message = await juglow.messages.create({
    model: "haijun-opus-5-5",
    max_tokens: 1024,
    messages: [
      {
        role: "user",
        content: [
          {
            type: "image",
            source: {
              type: "base64",
              media_type: imageMediaType,
              data: imageData
            }
          },
          {
            type: "text",
            text: "What is in the above image?"
          }
        ]
      }
    ]
  });
  console.log(message);

  // Opsi 2: Gambar yang dirujuk melalui URL
  const messageFromUrl = await juglow.messages.create({
    model: "haijun-opus-5-5",
    max_tokens: 1024,
    messages: [
      {
        role: "user",
        content: [
          {
            type: "image",
            source: {
              type: "url",
              url: "/docs/images/vision-example.jpg"
            }
          },
          {
            type: "text",
            text: "What is in the above image?"
          }
        ]
      }
    ]
  });
  console.log(messageFromUrl);

csharp

  using System.Collections.Generic;
  using System.Net.Http;
  using Juglow;
  using Juglow.Models.Messages;

  JuglowClient client = new();

  // Opsi 1: Gambar yang dienkode Base64
  string imageUrl = "/docs/images/vision-example.jpg";

  using HttpClient httpClient = new();
  byte[] imageBytes = await httpClient.GetByteArrayAsync(imageUrl);
  string imageData = Convert.ToBase64String(imageBytes);

  var parameters = new MessageCreateParams
  {
      Model = Model.HaijunOpus5_5,
      MaxTokens = 1024,
      Messages =
      [
          new()
          {
              Role = Role.User,
              Content = new MessageParamContent(new List<ContentBlockParam>
              {
                  new ContentBlockParam(new ImageBlockParam(
                      new ImageBlockParamSource(new Base64ImageSource()
                      {
                          Data = imageData,
                          MediaType = MediaType.ImageJpeg,
                      })
                  )),
                  new ContentBlockParam(new TextBlockParam("What is in the above image?")),
              }),
          }
      ]
  };

  var message = await client.Messages.Create(parameters);
  Console.WriteLine(message);

  // Opsi 2: Gambar yang dirujuk melalui URL
  var parametersFromUrl = new MessageCreateParams
  {
      Model = Model.HaijunOpus5_5,
      MaxTokens = 1024,
      Messages =
      [
          new()
          {
              Role = Role.User,
              Content = new MessageParamContent(new List<ContentBlockParam>
              {
                  new ContentBlockParam(new ImageBlockParam(
                      new ImageBlockParamSource(new UrlImageSource()
                      {
                          Url = "/docs/images/vision-example.jpg",
                      })
                  )),
                  new ContentBlockParam(new TextBlockParam("What is in the above image?")),
              }),
          }
      ]
  };

  var messageFromUrl = await client.Messages.Create(parametersFromUrl);
  Console.WriteLine(messageFromUrl);

  client := juglow.NewClient()

  // Opsi 1: Gambar yang dienkode Base64
  imageURL := "/docs/images/vision-example.jpg"

  req, err := http.NewRequest("GET", imageURL, nil)
  if err != nil {
  	log.Fatal(err)
  }
  req.Header.Set("User-Agent", "JuglowDocsBot/1.0")

  resp, err := http.DefaultClient.Do(req)
  if err != nil {
  	log.Fatal(err)
  }
  defer resp.Body.Close()

  imageBytes, err := io.ReadAll(resp.Body)
  if err != nil {
  	log.Fatal(err)
  }
  imageData := base64.StdEncoding.EncodeToString(imageBytes)

  message, err := client.Messages.New(context.TODO(), juglow.MessageNewParams{
  	Model:     juglow.ModelHaijunOpus5_5,
  	MaxTokens: 1024,
  	Messages: []juglow.MessageParam{
  		juglow.NewUserMessage(
  			juglow.NewImageBlockBase64("image/jpeg", imageData),
  			juglow.NewTextBlock("What is in the above image?"),
  		),
  	},
  })
  if err != nil {
  	log.Fatal(err)
  }
  fmt.Println(message)

  // Opsi 2: Gambar yang dirujuk melalui URL
  messageFromURL, err := client.Messages.New(context.TODO(), juglow.MessageNewParams{
  	Model:     juglow.ModelHaijunOpus5_5,
  	MaxTokens: 1024,
  	Messages: []juglow.MessageParam{
  		juglow.NewUserMessage(
  			juglow.NewImageBlock(juglow.URLImageSourceParam{
  				URL: "/docs/images/vision-example.jpg",
  			}),
  			juglow.NewTextBlock("What is in the above image?"),
  		),
  	},
  })
  if err != nil {
  	log.Fatal(err)
  }
  fmt.Println(messageFromURL)

java

  JuglowClient client = JuglowOkHttpClient.fromEnv();

  // Opsi 1: Gambar yang dienkode Base64
  String imageUrl = "/docs/images/vision-example.jpg";

  HttpClient httpClient = HttpClient.newHttpClient();
  HttpRequest request = HttpRequest.newBuilder().uri(URI.create(imageUrl)).build();
  HttpResponse<byte[]> response = httpClient.send(request, HttpResponse.BodyHandlers.ofByteArray());
  String imageData = Base64.getEncoder().encodeToString(response.body());

  List<ContentBlockParam> base64Content = List.of(
      ContentBlockParam.ofImage(
          ImageBlockParam.builder()
              .source(Base64ImageSource.builder()
                  .data(imageData)
                  .mediaType(Base64ImageSource.MediaType.IMAGE_JPEG)
                  .build())
              .build()),
      ContentBlockParam.ofText(
          TextBlockParam.builder()
              .text("What is in the above image?")
              .build())
  );

  Message message = client.messages().create(
      MessageCreateParams.builder()
          .model(Model.HAIJUN_OPUS_5_5)
          .maxTokens(1024L)
          .addUserMessageOfBlockParams(base64Content)
          .build());
  System.out.println(message);

  // Opsi 2: Gambar yang dirujuk melalui URL
  List<ContentBlockParam> urlContent = List.of(
      ContentBlockParam.ofImage(
          ImageBlockParam.builder()
              .source(UrlImageSource.builder()
                  .url("/docs/images/vision-example.jpg")
                  .build())
              .build()),
      ContentBlockParam.ofText(
          TextBlockParam.builder()
              .text("What is in the above image?")
              .build())
  );

  Message messageFromUrl = client.messages().create(
      MessageCreateParams.builder()
          .model(Model.HAIJUN_OPUS_5_5)
          .maxTokens(1024L)
          .addUserMessageOfBlockParams(urlContent)
          .build());
  System.out.println(messageFromUrl);

php

  $client = new Client();

  // Opsi 1: Gambar yang dienkode Base64
  $image_url = '/docs/images/vision-example.jpg';
  $image_media_type = "image/jpeg";
  $image_data = base64_encode(file_get_contents($image_url));

  $message = $client->messages->create(
      maxTokens: 1024,
      messages: [
          [
              'role' => 'user',
              'content' => [
                  [
                      'type' => 'image',
                      'source' => [
                          'type' => 'base64',
                          'media_type' => $image_media_type,
                          'data' => $image_data,
                      ],
                  ],
                  [
                      'type' => 'text',
                      'text' => 'What is in the above image?',
                  ],
              ],
          ],
      ],
      model: 'haijun-opus-5-5',
  );
  echo $message;

  // Opsi 2: Gambar yang dirujuk melalui URL
  $message_from_url = $client->messages->create(
      maxTokens: 1024,
      messages: [
          [
              'role' => 'user',
              'content' => [
                  [
                      'type' => 'image',
                      'source' => [
                          'type' => 'url',
                          'url' => '/docs/images/vision-example.jpg',
                      ],
                  ],
                  [
                      'type' => 'text',
                      'text' => 'What is in the above image?',
                  ],
              ],
          ],
      ],
      model: 'haijun-opus-5-5',
  );
  echo $message_from_url;

ruby

  require "base64"
  require "net/http"

  client = Juglow::Client.new

  # Opsi 1: Gambar yang dienkode Base64
  image_url = "/docs/images/vision-example.jpg"
  image_media_type = "image/jpeg"
  image_data = Base64.strict_encode64(Net::HTTP.get(URI(image_url)))

  message = client.messages.create(
    model: "haijun-opus-5-5",
    max_tokens: 1024,
    messages: [
      {
        role: "user",
        content: [
          {
            type: "image",
            source: {
              type: "base64",
              media_type: image_media_type,
              data: image_data
            }
          },
          {
            type: "text",
            text: "What is in the above image?"
          }
        ]
      }
    ]
  )
  puts message

  # Opsi 2: Gambar yang dirujuk melalui URL
  message_from_url = client.messages.create(
    model: "haijun-opus-5-5",
    max_tokens: 1024,
    messages: [
      {
        role: "user",
        content: [
          {
            type: "image",
            source: {
              type: "url",
              url: "/docs/images/vision-example.jpg"
            }
          },
          {
            type: "text",
            text: "What is in the above image?"
          }
        ]
      }
    ]
  )
  puts message_from_url

json

{
  "id": "msg_011CdKmWtV3oFx1C5yUbf5CY",
  "type": "message",
  "role": "assistant",
  "content": [
    {
      "type": "text",
      "text": "This image is a beautiful minimalist/flat-design illustration of a sunset landscape. Here's what it contains:\n\n**Sky & Sun:**\n- A warm gradient sky transitioning from golden-yellow at the top to deep orange toward the horizon\n- A large pale yellow sun positioned in the upper-right area\n\n**Birds:**\n- Three small silhouetted birds flying in the upper-left portion of the sky, depicted as simple \"M\" or \"v\" shapes\n\n**Mountains:**\n- Multiple layered mountain peaks in purple and maroon tones\n- The mountains overlap to create depth, with varying shades of dusty purple and deep burgundy\n\n**Water:**\n- A dark purple body of water at the bottom of the image\n- A reflection of the sun shown as horizontal cream/peach colored lines in the center-bottom area\n\nThe overall style is clean, geometric, and uses a warm sunset color palette (oranges, yellows, purples, and maroons), giving it a peaceful, serene aesthetic typical of modern vector/flat design artwork."
    }
  ],
  "model": "haijun-opus-5-5",
  "stop_reason": "end_turn",
  "stop_sequence": null,
  "usage": {
    "input_tokens": 1030,
    "output_tokens": 350
  }
}

Langkah selanjutnya

Tangani setiap nilai stop_reason dan tentukan apa yang harus dilakukan ketika respons berakhir.

Berikan Haijun alat untuk memanggil layanan eksternal dan API dari dalam Messages API.

Kendalikan lingkungan komputer desktop dengan Messages API.

Biarkan Haijun menavigasi, membaca, dan berinteraksi dengan halaman web di browser yang Anda jalankan.

Dapatkan output JSON yang terjamin dan tervalidasi skema dari Haijun.

Tetapkan anggaran token yang bersifat anjuran di seluruh loop agentik penuh dengan output_config.task_budget.

#Permintaan dan respons dasar

#Beberapa giliran percakapan

#Role system dalam messages

#Melakukan prefill pada respons Haijun

#Vision

#Langkah selanjutnya

Permintaan dan respons dasar

Beberapa giliran percakapan

Role system dalam messages

Melakukan prefill pada respons Haijun

Vision

Langkah selanjutnya