> 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/models.md).

# Modèles (liste)

Ce guide montre comment interroger **`GET /v1/models`** avec le SDK OpenAI (version “compatible OpenAI”) pour récupérer les modèles disponibles, puis filtrer par type (chat, embeddings, OCR, audio, etc.).

## Pré-requis

1. Définir votre clé API :

```bash
export ALBERT_API_KEY="votre_jeton"
```

2. Installer le SDK :

```bash
pip install -U openai
```

## Liste des modèles (`GET /v1/models`)

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

```bash
curl -sS "https://albert.api.etalab.gouv.fr/v1/models" \
  -H "Authorization: Bearer $ALBERT_API_KEY"
```

{% 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"],
)

models = client.models.list().data

for m in models[:3]:
    print(
        f"ID: {m.id}\n"
        f"Type: {m.type}\n"
        f"Owner: {m.owned_by}\n"
        f"Max tokens: {m.max_context_length}\n"
    )
```

{% 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 models = (await client.models.list()).data;
for (const m of models.slice(0, 3)) {
  console.log(`ID: ${m.id}`);
  console.log(`Type: ${m.type}`);
  console.log(`Owner: ${m.owned_by}`);
  console.log(`Max tokens: ${m.max_context_length}\n`);
}
```

{% endtab %}
{% endtabs %}

## Filtrer selon l’usage

En pratique, choisissez un modèle selon son champ `type` :

* `text-generation` : **chat** (`POST /v1/chat/completions`)
* `text-embeddings-inference` : **embeddings** (`POST /v1/embeddings`)
* `automatic-speech-recognition` : **audio** (`POST /v1/audio/transcriptions`)
* `image-to-text` ou variantes `*ocr*` : **OCR** (`POST /v1/ocr`)

{% hint style="warning" %}
⚠️ À vérifier — Le mapping exact entre `type` et les endpoints dépend de la configuration de votre instance. Utilisez la liste des modèles et vérifiez le comportement sur votre environnement.
{% endhint %}

## Vue catalogue (site officiel)

Pour une vue “humaine” des modèles (familles, licences, aliases), vous pouvez aussi consulter :

[Catalogue des modèles](https://albert.sites.beta.gouv.fr/solutions/models/)

Le site précise que les modèles proviennent de tiers (ex. Mistral, Meta) mais sont **hébergés sur nos serveurs** : aucune de vos données n’est envoyée à ces fournisseurs de modèles.

## Détail d’un modèle (`GET /v1/models/{model}`)

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

```bash
curl -sS "https://albert.api.etalab.gouv.fr/v1/models/REMPLACER_PAR_MODEL_ID" \
  -H "Authorization: Bearer $ALBERT_API_KEY"
```

{% endtab %}

{% tab title="Python" %}

```python
model_id = models[0].id

model = client.models.retrieve(model=model_id)

print(
    f"ID: {model.id}\n"
    f"Type: {model.type}\n"
    f"Owner: {model.owned_by}\n"
    f"Max tokens: {model.max_context_length}\n"
)
```

{% endtab %}

{% tab title="JavaScript" %}

```javascript
const modelId = models[0].id;
const model = await client.models.retrieve(modelId);

console.log(`ID: ${model.id}`);
console.log(`Type: ${model.type}`);
console.log(`Owner: ${model.owned_by}`);
console.log(`Max tokens: ${model.max_context_length}`);
```

{% endtab %}
{% endtabs %}

Pour la documentation endpoint, voir la [page “Models”](https://doc.incubateur.net/alliance/albert-api/api-reference/liste-des-endpoint/models).


---

# 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 dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:

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

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.

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.