> For the complete documentation index, see [llms.txt](https://guides.ia.numerique.gouv.fr/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://guides.ia.numerique.gouv.fr/albert-api/guides/streaming.md).

# Streaming

Pour le chat, le streaming s’active avec **`"stream": true`** sur **`POST /v1/chat/completions`**. Le serveur renvoie des **Server-Sent Events** (SSE) : chaque événement contient un fragment JSON (delta) ; le flux se termine par une ligne `data: [DONE]` comme sur l’API OpenAI.

## Paramètre `stream_options`

Lorsque `stream` est à `true`, vous pouvez renseigner **`stream_options`** pour affiner le comportement du flux (par exemple demander l’inclusion d’informations d’usage en fin de stream, sur le modèle des clients OpenAI).

```json
{
  "model": "votre-modele-text-generation",
  "messages": [{"role": "user", "content": "Bonjour"}],
  "stream": true,
  "stream_options": {
    "include_usage": true
  }
}
```

{% hint style="warning" %}
⚠️ À vérifier — Clés exactes supportées dans `stream_options` et compatibilité `include_usage` selon les modèles : consulter le schéma `CreateChatCompletion` dans la [page de l’endpoint Chat](https://doc.incubateur.net/alliance/albert-api/api-reference/liste-des-endpoint/chat).
{% endhint %}

## Exemples (curl / Python / JavaScript)

{% tabs %}
{% tab title="curl" %}

```bash
curl -N "https://albert.api.etalab.gouv.fr/v1/chat/completions" \
  -H "Authorization: Bearer $ALBERT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "votre-modele-text-generation",
    "messages": [{"role": "user", "content": "Raconte une phrase sur la météo."}],
    "stream": true
  }'
```

{% endtab %}

{% tab title="Python" %}

```python
import os
from openai import OpenAI

client = OpenAI(
    base_url="https://albert.api.etalab.gouv.fr/v1",
    api_key=os.environ["ALBERT_API_KEY"],
)

stream = client.chat.completions.create(
    model="votre-modele-text-generation",
    messages=[{"role": "user", "content": "Raconte une phrase sur la météo."}],
    stream=True,
)

for chunk in stream:
    delta = chunk.choices[0].delta
    if delta.content:
        print(delta.content, end="", flush=True)
print()
```

{% endtab %}

{% tab title="JavaScript" %}

```javascript
import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://albert.api.etalab.gouv.fr/v1",
  apiKey: process.env.ALBERT_API_KEY,
});

const stream = await client.chat.completions.create({
  model: "votre-modele-text-generation",
  messages: [{ role: "user", content: "Raconte une phrase sur la météo." }],
  stream: true,
});

for await (const chunk of stream) {
  const delta = chunk.choices?.[0]?.delta;
  if (delta?.content) process.stdout.write(delta.content);
}
process.stdout.write("\n");
```

{% endtab %}
{% endtabs %}

## Bonnes pratiques

* **Timeouts** : les streams longs exigent des timeouts côté client adaptés au contexte réseau.
* **Reconnexion** : en cas de coupure, la requête doit être relancée ; il n’y a pas de reprise automatique standard sur un identifiant de stream.
* **Erreurs mid-stream** : traiter les erreurs HTTP comme sur les appels non streamés ; voir [Codes d'erreur](broken://pages/N8WOLrEHNfTNle3o0aut).


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://guides.ia.numerique.gouv.fr/albert-api/guides/streaming.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.