---
title: Streaming de messages
url: https://platform.claude.com/docs/fr/build-with-claude/streaming
description: Diffusez les réponses de l'API Messages de manière incrémentale avec les server-sent events, y compris les deltas de texte, d'utilisation d'outils et de réflexion.
---

Lors de la création d'un Message, vous pouvez définir `"stream": true` pour diffuser la réponse de manière incrémentale en « streaming » (diffusion en flux) à l'aide de [« server-sent events »](https://developer.mozilla.org/en-US/Web/API/Server-sent%5Fevents/Using%5Fserver-sent%5Fevents) (événements envoyés par le serveur), ou SSE.

## Streaming avec les SDK

Le [SDK Python](https://github.com/anthropics/anthropic-sdk-python) et le [SDK TypeScript](https://github.com/anthropics/anthropic-sdk-typescript) offrent plusieurs façons d'effectuer du streaming. Le [SDK PHP](https://github.com/anthropics/anthropic-sdk-php) fournit le streaming via `createStream()`. Le SDK Python prend en charge les flux synchrones et asynchrones. Consultez la documentation de chaque SDK pour plus de détails.

<CodeGroup exclude="shell:cURL">
  ```bash CLI
  ant messages create --stream --format jsonl \
    --model claude-opus-5-5 \
    --max-tokens 1024 \
    --message '{role: user, content: "Hello"}' \
    | jq -rj 'select(.delta.type? == "text_delta") | .delta.text'
  ```

  ```python Python
  client = anthropic.Anthropic()

  with client.messages.stream(
      max_tokens=1024,
      messages=[{"role": "user", "content": "Hello"}],
      model="claude-opus-5-5",
  ) as stream:
      for text in stream.text_stream:
          print(text, end="", flush=True)
  ```

  ```typescript TypeScript
  const client = new Anthropic();

  await client.messages
    .stream({
      messages: [{ role: "user", content: "Hello" }],
      model: "claude-opus-5-5",
      max_tokens: 1024
    })
    .on("text", (text) => {
      console.log(text);
    });
  ```

  ```csharp C#
  AnthropicClient client = new();

  var parameters = new MessageCreateParams
  {
      Model = Model.ClaudeOpus5_5,
      MaxTokens = 1024,
      Messages = [new() { Role = Role.User, Content = "Hello" }]
  };

  await foreach (var msg in client.Messages.CreateStreaming(parameters))
  {
      Console.Write(msg);
  }
  ```

  ```go Go
  client := anthropic.NewClient()

  stream := client.Messages.NewStreaming(context.TODO(), anthropic.MessageNewParams{
  	Model:     anthropic.ModelClaudeOpus5_5,
  	MaxTokens: 1024,
  	Messages: []anthropic.MessageParam{
  		anthropic.NewUserMessage(anthropic.NewTextBlock("Hello")),
  	},
  })

  for stream.Next() {
  	event := stream.Current()
  	switch eventVariant := event.AsAny().(type) {
  	case anthropic.ContentBlockDeltaEvent:
  		switch deltaVariant := eventVariant.Delta.AsAny().(type) {
  		case anthropic.TextDelta:
  			fmt.Print(deltaVariant.Text)
  		}
  	}
  }
  if err := stream.Err(); err != nil {
  	log.Fatal(err)
  }
  ```

  ```java Java
  AnthropicClient client = AnthropicOkHttpClient.fromEnv();

  MessageCreateParams params = MessageCreateParams.builder()
      .model(Model.CLAUDE_OPUS_5_5)
      .maxTokens(1024L)
      .addUserMessage("Hello")
      .build();

  try (var streamResponse = client.messages().createStreaming(params)) {
      streamResponse.stream().forEach(event -> {
          event.contentBlockDelta().ifPresent(deltaEvent ->
              deltaEvent.delta().text().ifPresent(td ->
                  System.out.print(td.text())
              )
          );
      });
  }
  ```

  ```php PHP
  $client = new Client();

  $stream = $client->messages->createStream(
      maxTokens: 1024,
      messages: [
          ['role' => 'user', 'content' => 'Hello']
      ],
      model: 'claude-opus-5-5',
  );

  foreach ($stream as $message) {
      echo $message;
  }
  ```

  ```ruby Ruby
  client = Anthropic::Client.new

  stream = client.messages.stream(
    model: "claude-opus-5-5",
    max_tokens: 1024,
    messages: [{ role: "user", content: "Hello" }]
  )

  stream.text.each { |text| print(text) }
  ```
</CodeGroup>

## Obtenir le message final sans gérer les événements

Si vous n'avez pas besoin de traiter le texte au fur et à mesure de son arrivée, le SDK offre un moyen d'utiliser le streaming en interne tout en renvoyant l'objet `Message` complet, identique à ce que renvoie `client.messages.create()` (go: `client.Messages.New()`; java: `client.messages().create()`; csharp: `client.Messages.Create()`; php: `$client->messages->create()`). Ceci est particulièrement utile pour les requêtes avec de grandes valeurs de `max_tokens`, pour lesquelles le SDK exige le streaming afin d'éviter les délais d'expiration HTTP.

<CodeGroup exclude="shell:cURL">
  ```bash CLI
  # L'option --stream de la CLI ant émet un événement par ligne et ne les
  # accumule pas dans un Message final. Pour les longues générations, utilisez le streaming des
  # événements bruts :
  ant messages create --stream --format jsonl <<'YAML'
  model: claude-opus-5-5
  max_tokens: 128000
  messages:
    - role: user
      content: Write a detailed analysis...
  YAML
  ```

  ```python Python
  client = anthropic.Anthropic()

  with client.messages.stream(
      max_tokens=128000,
      messages=[{"role": "user", "content": "Write a detailed analysis..."}],
      model="claude-opus-5-5",
  ) as stream:
      message = stream.get_final_message()

  for block in message.content:
      if block.type == "text":
          print(block.text)
  ```

  ```typescript TypeScript
  const client = new Anthropic();

  const stream = client.messages.stream({
    max_tokens: 128000,
    messages: [{ role: "user", content: "Write a detailed analysis..." }],
    model: "claude-opus-5-5"
  });

  const message = await stream.finalMessage();
  const textBlock = message.content.find((block) => block.type === "text");
  if (textBlock && textBlock.type === "text") {
    console.log(textBlock.text);
  }
  ```

  ```csharp C#
  AnthropicClient client = new();

  var parameters = new MessageCreateParams
  {
      Model = Model.ClaudeOpus5_5,
      MaxTokens = 128000,
      Messages = [new() { Role = Role.User, Content = "Write a detailed analysis..." }]
  };

  // Pour traiter également les événements au fur et à mesure de leur arrivée, passez plutôt un MessageContentAggregator à CollectAsync().
  var message = await client.Messages.CreateStreaming(parameters).Aggregate();
  Console.WriteLine(message);
  ```

  ```go Go
  client := anthropic.NewClient()

  stream := client.Messages.NewStreaming(context.TODO(), anthropic.MessageNewParams{
  	Model:     anthropic.ModelClaudeOpus5_5,
  	MaxTokens: 128000,
  	Messages: []anthropic.MessageParam{
  		anthropic.NewUserMessage(anthropic.NewTextBlock("Write a detailed analysis...")),
  	},
  })

  message := anthropic.Message{}
  for stream.Next() {
  	event := stream.Current()
  	if err := message.Accumulate(event); err != nil {
  		log.Fatal(err)
  	}
  }
  if err := stream.Err(); err != nil {
  	log.Fatal(err)
  }

  for _, block := range message.Content {
  	if textBlock, ok := block.AsAny().(anthropic.TextBlock); ok {
  		fmt.Println(textBlock.Text)
  	}
  }
  ```

  ```java Java
  AnthropicClient client = AnthropicOkHttpClient.fromEnv();

  MessageCreateParams params = MessageCreateParams.builder()
      .model(Model.CLAUDE_OPUS_5_5)
      .maxTokens(128000L)
      .addUserMessage("Write a detailed analysis...")
      .build();

  MessageAccumulator accumulator = MessageAccumulator.create();
  try (var streamResponse = client.messages().createStreaming(params)) {
      streamResponse.stream().forEach(accumulator::accumulate);
  }

  Message message = accumulator.message();
  message.content().stream()
      .flatMap(block -> block.text().stream())
      .forEach(textBlock -> System.out.println(textBlock.text()));
  ```

  ```php PHP
  use Anthropic\Lib\Streaming\MessageAccumulator;

  $client = new Client();

  $stream = $client->messages->createStream(
      maxTokens: 128000,
      messages: [
          ['role' => 'user', 'content' => 'Write a detailed analysis...']
      ],
      model: 'claude-opus-5-5',
  );

  $accumulator = MessageAccumulator::forMessages();
  foreach ($stream as $event) {
      $accumulator->accumulate($event);
  }

  echo array_find($accumulator->message()->content, static fn ($block): bool => $block->type === 'text')->text;
  ```

  ```ruby Ruby
  client = Anthropic::Client.new

  message = client.messages.stream(
    model: "claude-opus-5-5",
    max_tokens: 128000,
    messages: [{ role: "user", content: "Write a detailed analysis..." }]
  ).accumulated_message

  message.content.each do |block|
    puts block.text if block.type == :text
  end
  ```
</CodeGroup>

L'appel `.stream()` (java: `.createStreaming()`; csharp: `.CreateStreaming()`; go: `.NewStreaming()`; php: `->createStream()`) maintient la connexion HTTP ouverte avec les server-sent events, puis l'utilitaire d'accumulation de messages du SDK, `stream.get_final_message()` (typescript: `stream.finalMessage()`; ruby: `stream.accumulated_message`; csharp: `.Aggregate()`; go: `message.Accumulate(event)`; java, php: `MessageAccumulator`), rassemble tous les événements dans l'objet `Message` complet.

## Types d'événements

Chaque événement envoyé par le serveur comprend un type d'événement nommé et des données JSON associées. Chaque événement utilise un nom d'événement SSE (par exemple, `event: message_stop`) et inclut le `type` d'événement correspondant dans ses données.

Chaque flux suit le déroulement d'événements suivant :

1. `message_start` : contient un objet `Message` avec un `content` vide. Avec l'en-tête bêta [`thinking-binding-controls-2026-08-01`](https://platform.claude.com/docs/fr/build-with-claude/thinking#preserved-thinking-controls), cet objet `Message` contient également le tableau `input_transformations`. Après un [repli côté serveur](https://platform.claude.com/docs/fr/build-with-claude/refusals-and-fallback#server-side-fallback) en cours de flux, l'événement `message_delta` final contient à nouveau le tableau avec les entrées du modèle ayant servi la réponse.
2. Une série de blocs de contenu, chacun ayant un événement `content_block_start`, un ou plusieurs événements `content_block_delta` et un événement `content_block_stop`. Chaque bloc de contenu possède un `index` qui correspond à son index dans le tableau `content` du Message final. Une exception : lors des réponses avec [repli côté serveur](https://platform.claude.com/docs/fr/build-with-claude/refusals-and-fallback#server-side-fallback), un bloc de contenu `fallback` arrive à chaque frontière de modèle sous la forme d'une paire `content_block_start` et `content_block_stop` sans aucun delta entre les deux.
3. Un ou plusieurs événements `message_delta`, indiquant des modifications de premier niveau de l'objet `Message` final.
4. Un événement `message_stop` final.

<Warning>
  Les décomptes de tokens affichés dans le champ `usage` de l'événement `message_delta` sont *cumulatifs*.
</Warning>

### Événements ping

Les flux d'événements peuvent également inclure un nombre quelconque d'événements `ping`.

### Événements d'erreur

L'API peut occasionnellement envoyer des [erreurs](https://platform.claude.com/docs/fr/api/errors) dans le flux d'événements. Par exemple, pendant les périodes de forte utilisation, vous pouvez recevoir une `overloaded_error`, qui correspondrait normalement à un HTTP 529 dans un contexte sans streaming :

```sse Example error
event: error
data: {"type": "error", "error": {"type": "overloaded_error", "message": "Overloaded"}}
```

### Autres événements

Conformément à la [politique de gestion des versions](https://platform.claude.com/docs/fr/api/versioning), de nouveaux types d'événements peuvent être ajoutés, et votre code doit gérer correctement les types d'événements inconnus.

## Types de deltas de blocs de contenu

Chaque événement `content_block_delta` contient un `delta` d'un type qui met à jour le bloc `content` à un `index` donné.

### Delta de texte

Un delta de bloc de contenu `text` ressemble à ceci :

```sse Text delta
event: content_block_delta
data: {"type": "content_block_delta","index": 0,"delta": {"type": "text_delta", "text": "ello frien"}}
```

### Delta JSON d'entrée

Les deltas des blocs de contenu `tool_use` correspondent à des mises à jour du champ `input` du bloc. Pour offrir une granularité maximale, les deltas sont des *chaînes JSON partielles*, tandis que le `tool_use.input` final est toujours un *objet*.

Vous pouvez accumuler les deltas de chaîne et analyser le JSON une fois que vous recevez un événement `content_block_stop`, en utilisant une bibliothèque comme [Pydantic](https://docs.pydantic.dev/latest/concepts/json/#partial-json-parsing) pour effectuer une analyse JSON partielle, ou en utilisant les [utilitaires de streaming](https://platform.claude.com/docs/fr/build-with-claude/streaming#streaming-with-sdks) du SDK. Dans les SDK Python et TypeScript, ces utilitaires vous donnent également accès à l'`input` analysé de manière incrémentale, au fur et à mesure du streaming.

Un delta de bloc de contenu `tool_use` ressemble à ceci :

```sse Input JSON delta
event: content_block_delta
data: {"type": "content_block_delta","index": 1,"delta": {"type": "input_json_delta","partial_json": "{\"location\": \"San Fra"}}}
```

Remarque : les modèles actuels ne prennent en charge que l'émission d'une seule propriété clé-valeur complète de `input` à la fois. Ainsi, lors de l'utilisation d'outils, il peut y avoir des délais entre les événements de streaming pendant que le modèle travaille. Une fois qu'une clé et une valeur de `input` sont accumulées, elles sont émises sous forme de plusieurs événements `content_block_delta` avec du JSON partiel découpé en morceaux, afin que le format puisse automatiquement prendre en charge une granularité plus fine dans les futurs modèles.

### Delta de réflexion

Lorsque vous utilisez la [réflexion](https://platform.claude.com/docs/fr/build-with-claude/thinking#streaming-thinking) avec le streaming activé, vous recevez le contenu de réflexion via des événements `thinking_delta`. Ces deltas correspondent au champ `thinking` des blocs de contenu `thinking`.

Pour le contenu de réflexion, un événement spécial `signature_delta` est envoyé juste avant l'événement `content_block_stop`. Cette signature est utilisée pour vérifier l'intégrité du bloc de réflexion.

Lorsque `display: "omitted"` est défini dans la configuration de réflexion, aucun texte de réflexion n'est diffusé. Le bloc de réflexion s'ouvre, reçoit un `thinking_delta` avec une chaîne `thinking` vide puis un unique `signature_delta`, et se ferme. Avec `display: "updates"` (bêta), les blocs de raisonnement sont diffusés de la même manière, et seules les [mises à jour de progression](https://platform.claude.com/docs/fr/build-with-claude/thinking#progress-updates) que certains modèles rédigent entre les appels d'outils diffusent des événements `thinking_delta` contenant du texte. Consultez [Contrôler l'affichage de la réflexion](https://platform.claude.com/docs/fr/build-with-claude/thinking#controlling-thinking-display).

Un delta de réflexion typique ressemble à ceci :

```sse Thinking delta
event: content_block_delta
data: {"type": "content_block_delta", "index": 0, "delta": {"type": "thinking_delta", "thinking": "I need to find the GCD of 1071 and 462 using the Euclidean algorithm.\n\n1071 = 2 × 462 + 147"}}
```

Le delta de signature ressemble à ceci :

```sse Signature delta
event: content_block_delta
data: {"type": "content_block_delta", "index": 0, "delta": {"type": "signature_delta", "signature": "EqQBCgIYAhIM1gbcDa9GJwZA2b3hGgxBdjrkzLoky3dl1pkiMOYds..."}}
```

## Réponse HTTP complète en streaming

Utilisez les [SDK clients](https://platform.claude.com/docs/fr/cli-sdks-libraries/overview) lorsque vous utilisez le mode streaming. Toutefois, si vous construisez une intégration directe avec l'API, vous devez gérer ces événements vous-même.

Une réponse en streaming se compose de :

1. Un événement `message_start`

2. Potentiellement plusieurs blocs de contenu, chacun contenant :

   * Un événement `content_block_start`
   * Potentiellement plusieurs événements `content_block_delta`
   * Un événement `content_block_stop`

3. Un ou plusieurs événements `message_delta`

4. Un événement `message_stop`

Des événements `ping` peuvent également être dispersés tout au long de la réponse. Consultez [Types d'événements](https://platform.claude.com/docs/fr/build-with-claude/streaming#event-types) pour plus de détails sur le format.

### Requête de streaming de base

<CodeGroup>
  ```bash cURL
  curl https://api.anthropic.com/v1/messages \
    -H "anthropic-version: 2023-06-01" \
    -H "content-type: application/json" \
    -H "x-api-key: $ANTHROPIC_API_KEY" \
    -d '{
      "model": "claude-opus-5-5",
      "messages": [{"role": "user", "content": "Hello"}],
      "max_tokens": 256,
      "stream": true
    }'
  ```

  ```bash CLI
  ant messages create --stream --format jsonl \
    --model claude-opus-5-5 \
    --max-tokens 256 \
    --message '{role: user, content: Hello}'
  ```

  ```python Python
  client = anthropic.Anthropic()

  with client.messages.stream(
      model="claude-opus-5-5",
      messages=[{"role": "user", "content": "Hello"}],
      max_tokens=256,
  ) as stream:
      for text in stream.text_stream:
          print(text, end="", flush=True)
  ```

  ```typescript TypeScript
  const client = new Anthropic();

  const stream = client.messages.stream({
    model: "claude-opus-5-5",
    messages: [{ role: "user", content: "Hello" }],
    max_tokens: 256
  });

  for await (const event of stream) {
    if (event.type === "content_block_delta" && event.delta.type === "text_delta") {
      process.stdout.write(event.delta.text);
    }
  }
  ```

  ```csharp C#
  AnthropicClient client = new();

  var parameters = new MessageCreateParams
  {
      Model = Model.ClaudeOpus5_5,
      MaxTokens = 256,
      Messages = [new() { Role = Role.User, Content = "Hello" }]
  };

  await foreach (var msg in client.Messages.CreateStreaming(parameters))
  {
      Console.Write(msg);
  }
  ```

  ```go Go
  client := anthropic.NewClient()

  stream := client.Messages.NewStreaming(context.TODO(), anthropic.MessageNewParams{
  	Model:     anthropic.ModelClaudeOpus5_5,
  	MaxTokens: 256,
  	Messages: []anthropic.MessageParam{
  		anthropic.NewUserMessage(anthropic.NewTextBlock("Hello")),
  	},
  })

  for stream.Next() {
  	event := stream.Current()
  	switch eventVariant := event.AsAny().(type) {
  	case anthropic.ContentBlockDeltaEvent:
  		switch deltaVariant := eventVariant.Delta.AsAny().(type) {
  		case anthropic.TextDelta:
  			fmt.Print(deltaVariant.Text)
  		}
  	}
  }
  if err := stream.Err(); err != nil {
  	log.Fatal(err)
  }
  ```

  ```java Java
  AnthropicClient client = AnthropicOkHttpClient.fromEnv();

  MessageCreateParams params = MessageCreateParams.builder()
      .model(Model.CLAUDE_OPUS_5_5)
      .maxTokens(256L)
      .addUserMessage("Hello")
      .build();

  try (var streamResponse = client.messages().createStreaming(params)) {
      streamResponse.stream().forEach(event -> {
          event.contentBlockDelta().ifPresent(deltaEvent ->
              deltaEvent.delta().text().ifPresent(td ->
                  System.out.print(td.text())
              )
          );
      });
  }
  ```

  ```php PHP
  $client = new Client();

  $stream = $client->messages->createStream(
      maxTokens: 256,
      messages: [
          ['role' => 'user', 'content' => 'Hello']
      ],
      model: 'claude-opus-5-5',
  );

  foreach ($stream as $message) {
      echo $message;
  }
  ```

  ```ruby Ruby
  client = Anthropic::Client.new

  stream = client.messages.stream(
    model: "claude-opus-5-5",
    messages: [{ role: "user", content: "Hello" }],
    max_tokens: 256
  )

  stream.text.each { |text| print(text) }
  ```
</CodeGroup>

```sse Response
event: message_start
data: {"type": "message_start", "message": {"id": "msg_1nZdL29xx5MUA1yADyHTEsnR8uuvGzszyY", "type": "message", "role": "assistant", "content": [], "model": "claude-opus-5-5", "stop_reason": null, "stop_sequence": null, "usage": {"input_tokens": 25, "output_tokens": 1}}}

event: content_block_start
data: {"type": "content_block_start", "index": 0, "content_block": {"type": "text", "text": ""}}

event: ping
data: {"type": "ping"}

event: content_block_delta
data: {"type": "content_block_delta", "index": 0, "delta": {"type": "text_delta", "text": "Hello"}}

event: content_block_delta
data: {"type": "content_block_delta", "index": 0, "delta": {"type": "text_delta", "text": "!"}}

event: content_block_stop
data: {"type": "content_block_stop", "index": 0}

event: message_delta
data: {"type": "message_delta", "delta": {"stop_reason": "end_turn", "stop_sequence":null}, "usage": {"output_tokens": 15}}

event: message_stop
data: {"type": "message_stop"}

```

### Requête de streaming avec utilisation d'outils

<Tip>
  L'utilisation d'outils prend en charge le [streaming à granularité fine](https://platform.claude.com/docs/fr/agents-and-tools/tool-use/fine-grained-tool-streaming) pour les valeurs de paramètres. Activez-le par outil avec `eager_input_streaming`.
</Tip>

Cette requête demande à Claude d'utiliser un outil pour indiquer la météo.

<CodeGroup>
  ```bash cURL
    curl https://api.anthropic.com/v1/messages \
      -H "content-type: application/json" \
      -H "x-api-key: $ANTHROPIC_API_KEY" \
      -H "anthropic-version: 2023-06-01" \
      -d '{
        "model": "claude-opus-5",
        "max_tokens": 1024,
        "tools": [
          {
            "name": "get_weather",
            "description": "Get the current weather in a given location",
            "input_schema": {
              "type": "object",
              "properties": {
                "location": {
                  "type": "string",
                  "description": "The city and state, e.g. San Francisco, CA"
                }
              },
              "required": ["location"]
            }
          }
        ],
        "tool_choice": {"type": "any"},
        "messages": [
          {
            "role": "user",
            "content": "What is the weather like in San Francisco?"
          }
        ],
        "stream": true
      }'
  ```

  ```bash CLI
  ant messages create --stream --format jsonl <<'YAML'
  model: claude-opus-5
  max_tokens: 1024
  tools:
    - name: get_weather
      description: Get the current weather in a given location
      input_schema:
        type: object
        properties:
          location:
            type: string
            description: The city and state, e.g. San Francisco, CA
        required:
          - location
  tool_choice:
    type: any
  messages:
    - role: user
      content: What is the weather like in San Francisco?
  YAML
  ```

  ```python Python
  client = anthropic.Anthropic()

  tools = [
      {
          "name": "get_weather",
          "description": "Get the current weather in a given location",
          "input_schema": {
              "type": "object",
              "properties": {
                  "location": {
                      "type": "string",
                      "description": "The city and state, e.g. San Francisco, CA",
                  }
              },
              "required": ["location"],
          },
      }
  ]

  with client.messages.stream(
      model="claude-opus-5",
      max_tokens=1024,
      tools=tools,
      tool_choice={"type": "any"},
      messages=[
          {"role": "user", "content": "What is the weather like in San Francisco?"}
      ],
  ) as stream:
      for text in stream.text_stream:
          print(text, end="", flush=True)
  ```

  ```typescript TypeScript
  const client = new Anthropic();

  const tools: Anthropic.Tool[] = [
    {
      name: "get_weather",
      description: "Get the current weather in a given location",
      input_schema: {
        type: "object",
        properties: {
          location: {
            type: "string",
            description: "The city and state, e.g. San Francisco, CA"
          }
        },
        required: ["location"]
      }
    }
  ];

  const stream = client.messages.stream({
    model: "claude-opus-5",
    max_tokens: 1024,
    tools: tools,
    tool_choice: { type: "any" },
    messages: [
      {
        role: "user",
        content: "What is the weather like in San Francisco?"
      }
    ]
  });

  for await (const event of stream) {
    if (event.type === "content_block_delta" && event.delta.type === "text_delta") {
      process.stdout.write(event.delta.text);
    }
  }
  ```

  ```csharp C#
  AnthropicClient client = new();

  var parameters = new MessageCreateParams
  {
      Model = Model.ClaudeOpus5,
      MaxTokens = 1024,
      Tools = [
          new ToolUnion(new Tool()
          {
              Name = "get_weather",
              Description = "Get the current weather in a given location",
              InputSchema = new InputSchema()
              {
                  Properties = new Dictionary<string, JsonElement>
                  {
                      ["location"] = JsonSerializer.SerializeToElement(new { type = "string", description = "The city and state, e.g. San Francisco, CA" }),
                  },
                  Required = ["location"],
              },
          }),
      ],
      ToolChoice = new ToolChoiceAny(),
      Messages = [
          new() { Role = Role.User, Content = "What is the weather like in San Francisco?" }
      ]
  };

  await foreach (var msg in client.Messages.CreateStreaming(parameters))
  {
      Console.Write(msg);
  }
  ```

  ```go Go
  client := anthropic.NewClient()

  stream := client.Messages.NewStreaming(context.TODO(), anthropic.MessageNewParams{
  	Model:     anthropic.ModelClaudeOpus5,
  	MaxTokens: 1024,
  	Tools: []anthropic.ToolUnionParam{
  		{OfTool: &anthropic.ToolParam{
  			Name:        "get_weather",
  			Description: anthropic.String("Get the current weather in a given location"),
  			InputSchema: anthropic.ToolInputSchemaParam{
  				Properties: map[string]any{
  					"location": map[string]any{
  						"type":        "string",
  						"description": "The city and state, e.g. San Francisco, CA",
  					},
  				},
  				Required: []string{"location"},
  			},
  		}},
  	},
  	ToolChoice: anthropic.ToolChoiceUnionParam{OfAny: &anthropic.ToolChoiceAnyParam{}},
  	Messages: []anthropic.MessageParam{
  		anthropic.NewUserMessage(anthropic.NewTextBlock("What is the weather like in San Francisco?")),
  	},
  })

  for stream.Next() {
  	event := stream.Current()
  	switch eventVariant := event.AsAny().(type) {
  	case anthropic.ContentBlockDeltaEvent:
  		switch deltaVariant := eventVariant.Delta.AsAny().(type) {
  		case anthropic.TextDelta:
  			fmt.Print(deltaVariant.Text)
  		}
  	}
  }
  if err := stream.Err(); err != nil {
  	log.Fatal(err)
  }
  ```

  ```java Java
  AnthropicClient client = AnthropicOkHttpClient.fromEnv();

  MessageCreateParams params = MessageCreateParams.builder()
      .model(Model.CLAUDE_OPUS_5)
      .maxTokens(1024L)
      .addTool(Tool.builder()
          .name("get_weather")
          .description("Get the current weather in a given location")
          .inputSchema(Tool.InputSchema.builder()
              .properties(JsonValue.from(Map.of(
                  "location", Map.of(
                      "type", "string",
                      "description", "The city and state, e.g. San Francisco, CA"
                  )
              )))
              .putAdditionalProperty("required", JsonValue.from(List.of("location")))
              .build())
          .build())
      .toolChoice(ToolChoice.ofAny(ToolChoiceAny.builder().build()))
      .addUserMessage("What is the weather like in San Francisco?")
      .build();

  try (var streamResponse = client.messages().createStreaming(params)) {
      streamResponse.stream().forEach(event -> {
          event.contentBlockDelta().ifPresent(deltaEvent ->
              deltaEvent.delta().text().ifPresent(td ->
                  System.out.print(td.text())
              )
          );
      });
  }
  ```

  ```php PHP
  $client = new Client();

  $stream = $client->messages->createStream(
      maxTokens: 1024,
      messages: [
          ['role' => 'user', 'content' => 'What is the weather like in San Francisco?']
      ],
      model: 'claude-opus-5',
      toolChoice: ['type' => 'any'],
      tools: [
          [
              'name' => 'get_weather',
              'description' => 'Get the current weather in a given location',
              'input_schema' => [
                  'type' => 'object',
                  'properties' => [
                      'location' => [
                          'type' => 'string',
                          'description' => 'The city and state, e.g. San Francisco, CA'
                      ]
                  ],
                  'required' => ['location']
              ]
          ]
      ],
  );

  foreach ($stream as $message) {
      echo $message;
  }
  ```

  ```ruby Ruby
  client = Anthropic::Client.new

  tools = [
    {
      name: "get_weather",
      description: "Get the current weather in a given location",
      input_schema: {
        type: "object",
        properties: {
          location: {
            type: "string",
            description: "The city and state, e.g. San Francisco, CA"
          }
        },
        required: ["location"]
      }
    }
  ]

  stream = client.messages.stream(
    model: "claude-opus-5",
    max_tokens: 1024,
    tools: tools,
    tool_choice: { type: "any" },
    messages: [
      { role: "user", content: "What is the weather like in San Francisco?" }
    ]
  )

  stream.text.each { |text| print(text) }
  ```
</CodeGroup>

```sse Response
event: message_start
data: {"type":"message_start","message":{"id":"msg_014p7gG3wDgGV9EUtLvnow3U","type":"message","role":"assistant","model":"claude-opus-5","stop_sequence":null,"usage":{"input_tokens":472,"output_tokens":2},"content":[],"stop_reason":null}}

event: content_block_start
data: {"type":"content_block_start","index":0,"content_block":{"type":"text","text":""}}

event: ping
data: {"type": "ping"}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":"Okay"}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":","}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":" let"}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":"'s"}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":" check"}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":" the"}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":" weather"}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":" for"}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":" San"}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":" Francisco"}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":","}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":" CA"}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":":"}}

event: content_block_stop
data: {"type":"content_block_stop","index":0}

event: content_block_start
data: {"type":"content_block_start","index":1,"content_block":{"type":"tool_use","id":"toolu_01T1x1fJ34qAmk2tNTrN7Up6","name":"get_weather","input":{}}}

event: content_block_delta
data: {"type":"content_block_delta","index":1,"delta":{"type":"input_json_delta","partial_json":""}}

event: content_block_delta
data: {"type":"content_block_delta","index":1,"delta":{"type":"input_json_delta","partial_json":"{\"location\":"}}

event: content_block_delta
data: {"type":"content_block_delta","index":1,"delta":{"type":"input_json_delta","partial_json":" \"San"}}

event: content_block_delta
data: {"type":"content_block_delta","index":1,"delta":{"type":"input_json_delta","partial_json":" Francisc"}}

event: content_block_delta
data: {"type":"content_block_delta","index":1,"delta":{"type":"input_json_delta","partial_json":"o,"}}

event: content_block_delta
data: {"type":"content_block_delta","index":1,"delta":{"type":"input_json_delta","partial_json":" CA\"}"}}

event: content_block_stop
data: {"type":"content_block_stop","index":1}

event: message_delta
data: {"type":"message_delta","delta":{"stop_reason":"tool_use","stop_sequence":null},"usage":{"output_tokens":89}}

event: message_stop
data: {"type":"message_stop"}
```

### Requête de streaming avec réflexion

Cette requête active la réflexion avec le streaming. Le paramètre `display: "summarized"` transmet en streaming un résumé condensé du raisonnement de Claude plutôt que la chaîne de pensée complète.

<CodeGroup>
  ```bash cURL
  curl https://api.anthropic.com/v1/messages \
    -H "x-api-key: $ANTHROPIC_API_KEY" \
    -H "anthropic-version: 2023-06-01" \
    -H "content-type: application/json" \
    -d '{
      "model": "claude-opus-5-5",
      "max_tokens": 20000,
      "stream": true,
      "thinking": {
        "type": "adaptive",
        "display": "summarized"
      },
      "messages": [
        {
          "role": "user",
          "content": "What is the greatest common divisor of 1071 and 462?"
        }
      ]
    }'
  ```

  ```bash CLI
  ant messages create --stream --format jsonl \
    --model claude-opus-5-5 \
    --max-tokens 20000 \
    --thinking '{type: adaptive, display: summarized}' \
    --message '{role: user, content: What is the greatest common divisor of 1071 and 462?}'
  ```

  ```python Python
  client = anthropic.Anthropic()

  with client.messages.stream(
      model="claude-opus-5-5",
      max_tokens=20000,
      thinking={"type": "adaptive", "display": "summarized"},
      messages=[
          {
              "role": "user",
              "content": "What is the greatest common divisor of 1071 and 462?",
          }
      ],
  ) as stream:
      for event in stream:
          if event.type == "content_block_delta":
              delta = event.delta
              match delta.type:
                  case "thinking_delta":
                      print(delta.thinking, end="", flush=True)
                  case "text_delta":
                      print(delta.text, end="", flush=True)
  ```

  ```typescript TypeScript
  const client = new Anthropic();

  const stream = client.messages.stream({
    model: "claude-opus-5-5",
    max_tokens: 20000,
    thinking: { type: "adaptive", display: "summarized" },
    messages: [
      {
        role: "user",
        content: "What is the greatest common divisor of 1071 and 462?"
      }
    ]
  });

  for await (const event of stream) {
    if (event.type === "content_block_delta") {
      switch (event.delta.type) {
        case "thinking_delta":
          process.stdout.write(event.delta.thinking);
          break;
        case "text_delta":
          process.stdout.write(event.delta.text);
          break;
      }
    }
  }
  ```

  ```csharp C#
  using Anthropic;
  using Anthropic.Models.Messages;

  AnthropicClient client = new();

  var parameters = new MessageCreateParams
  {
      Model = Model.ClaudeOpus5_5,
      MaxTokens = 20000,
      Thinking = new ThinkingConfigAdaptive { Display = Display.Summarized },
      Messages = [new() { Role = Role.User, Content = "What is the greatest common divisor of 1071 and 462?" }]
  };

  await foreach (var msg in client.Messages.CreateStreaming(parameters))
  {
      Console.Write(msg);
  }
  ```

  ```go Go
  client := anthropic.NewClient()

  stream := client.Messages.NewStreaming(context.TODO(), anthropic.MessageNewParams{
  	Model:     anthropic.ModelClaudeOpus5_5,
  	MaxTokens: 20000,
  	Thinking: anthropic.ThinkingConfigParamUnion{
  		OfAdaptive: &anthropic.ThinkingConfigAdaptiveParam{
  			Display: anthropic.ThinkingConfigAdaptiveDisplaySummarized,
  		},
  	},
  	Messages: []anthropic.MessageParam{
  		anthropic.NewUserMessage(anthropic.NewTextBlock("What is the greatest common divisor of 1071 and 462?")),
  	},
  })

  for stream.Next() {
  	event := stream.Current()
  	switch eventVariant := event.AsAny().(type) {
  	case anthropic.ContentBlockDeltaEvent:
  		switch deltaVariant := eventVariant.Delta.AsAny().(type) {
  		case anthropic.ThinkingDelta:
  			fmt.Print(deltaVariant.Thinking)
  		case anthropic.TextDelta:
  			fmt.Print(deltaVariant.Text)
  		}
  	}
  }
  if err := stream.Err(); err != nil {
  	log.Fatal(err)
  }
  ```

  ```java Java
  AnthropicClient client = AnthropicOkHttpClient.fromEnv();

  MessageCreateParams params = MessageCreateParams.builder()
      .model(Model.CLAUDE_OPUS_5_5)
      .maxTokens(20000L)
      .thinking(ThinkingConfigAdaptive.builder()
          .display(ThinkingConfigAdaptive.Display.SUMMARIZED)
          .build())
      .addUserMessage("What is the greatest common divisor of 1071 and 462?")
      .build();

  try (var streamResponse = client.messages().createStreaming(params)) {
      streamResponse.stream().forEach(event -> {
          event.contentBlockDelta().ifPresent(deltaEvent -> {
              deltaEvent.delta().thinking().ifPresent(td ->
                  IO.print(td.thinking())
              );
              deltaEvent.delta().text().ifPresent(td ->
                  IO.print(td.text())
              );
          });
      });
  }
  ```

  ```php PHP
  $client = new Client();

  $stream = $client->messages->createStream(
      maxTokens: 20000,
      messages: [
          ['role' => 'user', 'content' => 'What is the greatest common divisor of 1071 and 462?']
      ],
      model: 'claude-opus-5-5',
      thinking: ['type' => 'adaptive', 'display' => 'summarized'],
  );

  foreach ($stream as $message) {
      echo $message;
  }
  ```

  ```ruby Ruby
  client = Anthropic::Client.new

  stream = client.messages.stream(
    model: "claude-opus-5-5",
    max_tokens: 20000,
    thinking: { type: "adaptive", display: "summarized" },
    messages: [
      { role: "user", content: "What is the greatest common divisor of 1071 and 462?" }
    ]
  )

  stream.each do |event|
    if event.is_a?(Anthropic::Models::RawContentBlockDeltaEvent)
      delta = event.delta
      case delta
      when Anthropic::Models::ThinkingDelta
        print(delta.thinking)
      when Anthropic::Models::TextDelta
        print(delta.text)
      end
    end
  end
  ```
</CodeGroup>

```sse Response
event: message_start
data: {"type": "message_start", "message": {"id": "msg_01...", "type": "message", "role": "assistant", "content": [], "model": "claude-opus-5-5", "stop_reason": null, "stop_sequence": null}}

event: content_block_start
data: {"type": "content_block_start", "index": 0, "content_block": {"type": "thinking", "thinking": "", "signature": ""}}

event: content_block_delta
data: {"type": "content_block_delta", "index": 0, "delta": {"type": "thinking_delta", "thinking": "I need to find the GCD of 1071 and 462 using the Euclidean algorithm.\n\n1071 = 2 × 462 + 147"}}

event: content_block_delta
data: {"type": "content_block_delta", "index": 0, "delta": {"type": "thinking_delta", "thinking": "\n462 = 3 × 147 + 21"}}

event: content_block_delta
data: {"type": "content_block_delta", "index": 0, "delta": {"type": "thinking_delta", "thinking": "\n147 = 7 × 21 + 0"}}

event: content_block_delta
data: {"type": "content_block_delta", "index": 0, "delta": {"type": "thinking_delta", "thinking": "\nThe remainder is 0, so GCD(1071, 462) = 21."}}

event: content_block_delta
data: {"type": "content_block_delta", "index": 0, "delta": {"type": "signature_delta", "signature": "EqQBCgIYAhIM1gbcDa9GJwZA2b3hGgxBdjrkzLoky3dl1pkiMOYds..."}}

event: content_block_stop
data: {"type": "content_block_stop", "index": 0}

event: content_block_start
data: {"type": "content_block_start", "index": 1, "content_block": {"type": "text", "text": ""}}

event: content_block_delta
data: {"type": "content_block_delta", "index": 1, "delta": {"type": "text_delta", "text": "The greatest common divisor of 1071 and 462 is **21**."}}

event: content_block_stop
data: {"type": "content_block_stop", "index": 1}

event: message_delta
data: {"type": "message_delta", "delta": {"stop_reason": "end_turn", "stop_sequence": null}}

event: message_stop
data: {"type": "message_stop"}
```

### Requête de streaming avec utilisation de l'outil de recherche web

Cette requête demande à Claude de rechercher sur le web des informations météorologiques actuelles.

<CodeGroup>
  ```bash cURL
  curl https://api.anthropic.com/v1/messages \
    -H "x-api-key: $ANTHROPIC_API_KEY" \
    -H "anthropic-version: 2023-06-01" \
    -H "content-type: application/json" \
    -d '{
      "model": "claude-opus-5-5",
      "max_tokens": 1024,
      "stream": true,
      "tools": [
        {
          "type": "web_search_20250305",
          "name": "web_search",
          "max_uses": 5
        }
      ],
      "messages": [
        {
          "role": "user",
          "content": "What is the weather like in New York City today?"
        }
      ]
    }'
  ```

  ```bash CLI
  ant messages create --stream --format jsonl \
    --model claude-opus-5-5 \
    --max-tokens 1024 \
    --tool '{type: web_search_20250305, name: web_search, max_uses: 5}' \
    --message '{role: user, content: What is the weather like in New York City today?}'
  ```

  ```python Python
  client = anthropic.Anthropic()

  with client.messages.stream(
      model="claude-opus-5-5",
      max_tokens=1024,
      tools=[{"type": "web_search_20250305", "name": "web_search", "max_uses": 5}],
      messages=[
          {"role": "user", "content": "What is the weather like in New York City today?"}
      ],
  ) as stream:
      for text in stream.text_stream:
          print(text, end="", flush=True)
  ```

  ```typescript TypeScript
  const client = new Anthropic();

  const stream = client.messages.stream({
    model: "claude-opus-5-5",
    max_tokens: 1024,
    tools: [{ type: "web_search_20250305", name: "web_search", max_uses: 5 }],
    messages: [{ role: "user", content: "What is the weather like in New York City today?" }]
  });

  for await (const event of stream) {
    if (event.type === "content_block_delta" && event.delta.type === "text_delta") {
      process.stdout.write(event.delta.text);
    }
  }
  ```

  ```csharp C#
  using Anthropic;
  using Anthropic.Models.Messages;

  AnthropicClient client = new();

  var parameters = new MessageCreateParams
  {
      Model = Model.ClaudeOpus5_5,
      MaxTokens = 1024,
      Tools = [new ToolUnion(new WebSearchTool20250305() { MaxUses = 5 })],
      Messages = [new() { Role = Role.User, Content = "What is the weather like in New York City today?" }]
  };

  await foreach (var msg in client.Messages.CreateStreaming(parameters))
  {
      Console.Write(msg);
  }
  ```

  ```go Go
  client := anthropic.NewClient()

  stream := client.Messages.NewStreaming(context.TODO(), anthropic.MessageNewParams{
  	Model:     anthropic.ModelClaudeOpus5_5,
  	MaxTokens: 1024,
  	Tools: []anthropic.ToolUnionParam{
  		{
  			OfWebSearchTool20250305: &anthropic.WebSearchTool20250305Param{
  				MaxUses: anthropic.Int(5),
  			},
  		},
  	},
  	Messages: []anthropic.MessageParam{
  		anthropic.NewUserMessage(anthropic.NewTextBlock("What is the weather like in New York City today?")),
  	},
  })

  for stream.Next() {
  	event := stream.Current()
  	switch eventVariant := event.AsAny().(type) {
  	case anthropic.ContentBlockDeltaEvent:
  		switch deltaVariant := eventVariant.Delta.AsAny().(type) {
  		case anthropic.TextDelta:
  			fmt.Print(deltaVariant.Text)
  		}
  	}
  }
  if err := stream.Err(); err != nil {
  	log.Fatal(err)
  }
  ```

  ```java Java
  AnthropicClient client = AnthropicOkHttpClient.fromEnv();

  MessageCreateParams params = MessageCreateParams.builder()
      .model(Model.CLAUDE_OPUS_5_5)
      .maxTokens(1024L)
      .addTool(WebSearchTool20250305.builder()
          .maxUses(5L)
          .build())
      .addUserMessage("What is the weather like in New York City today?")
      .build();

  try (var streamResponse = client.messages().createStreaming(params)) {
      streamResponse.stream().forEach(event -> {
          event.contentBlockDelta().ifPresent(deltaEvent ->
              deltaEvent.delta().text().ifPresent(td ->
                  System.out.print(td.text())
              )
          );
      });
  }
  ```

  ```php PHP
  $client = new Client();

  $stream = $client->messages->createStream(
      maxTokens: 1024,
      messages: [
          ['role' => 'user', 'content' => 'What is the weather like in New York City today?']
      ],
      model: 'claude-opus-5-5',
      tools: [
          ['type' => 'web_search_20250305', 'name' => 'web_search', 'max_uses' => 5]
      ],
  );

  foreach ($stream as $message) {
      echo $message;
  }
  ```

  ```ruby Ruby
  client = Anthropic::Client.new

  stream = client.messages.stream(
    model: :"claude-opus-5-5",
    max_tokens: 1024,
    tools: [
      {
        type: "web_search_20250305",
        name: "web_search",
        max_uses: 5
      }
    ],
    messages: [
      {
        role: "user",
        content: "What is the weather like in New York City today?"
      }
    ]
  )

  stream.text.each { |text| print(text) }
  ```
</CodeGroup>

```sse Response
event: message_start
data: {"type":"message_start","message":{"id":"msg_01G...","type":"message","role":"assistant","model":"claude-opus-5-5","content":[],"stop_reason":null,"stop_sequence":null,"usage":{"input_tokens":2679,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"output_tokens":3}}}

event: content_block_start
data: {"type":"content_block_start","index":0,"content_block":{"type":"text","text":""}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":"I'll check"}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":" the current weather in New York City for you"}}

event: ping
data: {"type": "ping"}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":"."}}

event: content_block_stop
data: {"type":"content_block_stop","index":0}

event: content_block_start
data: {"type":"content_block_start","index":1,"content_block":{"type":"server_tool_use","id":"srvtoolu_014hJH82Qum7Td6UV8gDXThB","name":"web_search","input":{}}}

event: content_block_delta
data: {"type":"content_block_delta","index":1,"delta":{"type":"input_json_delta","partial_json":""}}

event: content_block_delta
data: {"type":"content_block_delta","index":1,"delta":{"type":"input_json_delta","partial_json":"{\"query"}}

event: content_block_delta
data: {"type":"content_block_delta","index":1,"delta":{"type":"input_json_delta","partial_json":"\":"}}

event: content_block_delta
data: {"type":"content_block_delta","index":1,"delta":{"type":"input_json_delta","partial_json":" \"weather"}}

event: content_block_delta
data: {"type":"content_block_delta","index":1,"delta":{"type":"input_json_delta","partial_json":" NY"}}

event: content_block_delta
data: {"type":"content_block_delta","index":1,"delta":{"type":"input_json_delta","partial_json":"C to"}}

event: content_block_delta
data: {"type":"content_block_delta","index":1,"delta":{"type":"input_json_delta","partial_json":"day\"}"}}

event: content_block_stop
data: {"type":"content_block_stop","index":1 }

event: content_block_start
data: {"type":"content_block_start","index":2,"content_block":{"type":"web_search_tool_result","tool_use_id":"srvtoolu_014hJH82Qum7Td6UV8gDXThB","content":[{"type":"web_search_result","title":"Weather in New York City in May 2025 (New York) - detailed Weather Forecast for a month","url":"https://world-weather.info/forecast/usa/new_york/may-2025/","encrypted_content":"Ev0DCioIAxgCIiQ3NmU4ZmI4OC1k...","page_age":null},...]}}

event: content_block_stop
data: {"type":"content_block_stop","index":2}

event: content_block_start
data: {"type":"content_block_start","index":3,"content_block":{"type":"text","text":""}}

event: content_block_delta
data: {"type":"content_block_delta","index":3,"delta":{"type":"text_delta","text":"Here's the current weather information for New York"}}

event: content_block_delta
data: {"type":"content_block_delta","index":3,"delta":{"type":"text_delta","text":" City:\n\n# Weather"}}

event: content_block_delta
data: {"type":"content_block_delta","index":3,"delta":{"type":"text_delta","text":" in New York City"}}

event: content_block_delta
data: {"type":"content_block_delta","index":3,"delta":{"type":"text_delta","text":"\n\n"}}

...

event: content_block_stop
data: {"type":"content_block_stop","index":17}

event: message_delta
data: {"type":"message_delta","delta":{"stop_reason":"end_turn","stop_sequence":null},"usage":{"input_tokens":10682,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"output_tokens":510,"server_tool_use":{"web_search_requests":1}}}

event: message_stop
data: {"type":"message_stop"}
```

## Récupération après erreur

### Claude 4.5 et versions antérieures

Pour les modèles Claude 4.5 et antérieurs, vous pouvez récupérer une requête de streaming interrompue en raison de problèmes réseau, de délais d'expiration ou d'autres erreurs en reprenant à partir de l'endroit où le flux a été interrompu. Cette approche vous évite de retraiter l'intégralité de la réponse.

La stratégie de récupération de base consiste à :

1. **Capturer la réponse partielle :** enregistrez tout le contenu reçu avec succès avant que l'erreur ne se produise.
2. **Construire une requête de continuation :** créez une nouvelle requête API qui inclut la réponse partielle de l'assistant comme début d'un nouveau message d'assistant.
3. **Reprendre le streaming :** continuez à recevoir le reste de la réponse à partir de l'endroit où elle a été interrompue.

### Claude 4.6 et versions ultérieures

Pour les modèles Claude 4.6 et ultérieurs, la même stratégie de capture et de reprise s'applique, mais l'étape 2 change : au lieu de placer la réponse partielle dans un message d'assistant, ajoutez un message utilisateur qui demande au modèle de continuer là où il s'est arrêté.

1. **Capturer la réponse partielle :** enregistrez tout le contenu reçu avec succès avant que l'erreur ne se produise.
2. **Construire une requête de continuation :** créez une nouvelle requête API avec un message utilisateur contenant la réponse partielle et une instruction de continuer, par exemple :
   ```text Sample prompt wrap
   Your previous response was interrupted and ended with [previous_response]. Continue from where you left off.
   ```
3. **Reprendre le streaming :** continuez à recevoir le reste de la réponse à partir de l'endroit où elle a été interrompue.

### Bonnes pratiques de récupération après erreur

1. **Utilisez les fonctionnalités du SDK :** tirez parti des capacités intégrées d'accumulation de messages et de gestion des erreurs du SDK.
2. **Gérez les types de contenu :** sachez que les messages peuvent contenir plusieurs blocs de contenu (`text`, `tool_use`, `thinking`). Les blocs d'utilisation d'outils et de réflexion ne peuvent pas être récupérés partiellement. Vous pouvez reprendre le streaming à partir du bloc de texte le plus récent.

## Étapes suivantes

<CardGroup cols={2}>
  <Card title="Raisons d'arrêt et repli" icon="list" href="https://platform.claude.com/docs/fr/build-with-claude/handling-stop-reasons">
    Gérez chaque valeur de `stop_reason` une fois qu'un flux est terminé.
  </Card>

  <Card title="Streaming d'outils à granularité fine" icon="wrench" href="https://platform.claude.com/docs/fr/agents-and-tools/tool-use/fine-grained-tool-streaming">
    Diffusez le JSON d'entrée des outils sans mise en mémoire tampon côté serveur pour une latence réduite.
  </Card>

  <Card title="Réflexion" icon="brain" href="https://platform.claude.com/docs/fr/build-with-claude/thinking">
    Diffusez la sortie de réflexion avec les événements `thinking_delta` et `signature_delta`.
  </Card>

  <Card title="SDK clients" icon="code" href="https://platform.claude.com/docs/fr/cli-sdks-libraries/overview">
    Utilisez les SDK officiels, qui gèrent pour vous le streaming, l'accumulation et la reconnexion.
  </Card>

  <Card title="Traitement par lots" icon="stack" href="https://platform.claude.com/docs/fr/build-with-claude/batch-processing">
    Traitez de grands volumes de requêtes de manière asynchrone lorsque vous n'avez pas besoin de réponses en temps réel.
  </Card>
</CardGroup>