> ## Documentation Index
> Fetch the complete documentation index at: https://docs.atako.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Mistral AI

> Connect Mistral AI to your Atako agents — 2 read and 4 write actions.

Let your agents use the Mistral AI API — chat completions, embeddings, OCR, moderation and the model catalogue.

## Connection

* **Authentication**: API key.

<Note>
  console.mistral.ai → sign in → "API Keys" in the left sidebar → "Create new key". Add a Name, set an Expiration date, choose a Connector access scope, then click "Create new key" and copy the value immediately — it is shown only once. A standard workspace API key covers chat completions, embeddings, OCR, moderation and models.

  See [Mistral AI's documentation](https://console.mistral.ai/api-keys).
</Note>

## Read actions (2)

| Action        | Description                               |
| ------------- | ----------------------------------------- |
| `get_model`   | Fetch metadata for a single model by id.  |
| `list_models` | List the models available to the API key. |

## Write actions (4)

| Action                | Description                                                                                                                                                                                                                                                                                                                                                                                                                                          |
| --------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `chat_completion`     | Generate a model reply for a list of chat messages (non-streaming). Arguments: model (string, e.g. "mistral-large-latest"); messages (array of objects \{role: "system"\|"user"\|"assistant"\|"tool", content: string}); temperature (number 0–1.5); top\_p (number > 0 and ≤ 1); max\_tokens (integer ≥ 1); stop (array of strings); random\_seed (integer ≥ 0); response\_format (object \{type: "text"\|"json\_object"}); safe\_prompt (boolean). |
| `classify_moderation` | Classify raw text against Mistral content-safety categories. Arguments: model (string, e.g. "mistral-moderation-latest"); input (a single string, or an array of strings).                                                                                                                                                                                                                                                                           |
| `embeddings`          | Compute embedding vectors for one or more input texts. Arguments: model (string, e.g. "mistral-embed"); input (a single string, or an array of strings); output\_dimension (integer ≥ 1); encoding\_format ("float" or "base64").                                                                                                                                                                                                                    |
| `ocr`                 | Extract text and structure from a document or an image reachable by URL. Arguments: model (string, e.g. "mistral-ocr-latest"); document (object — either \{type: "document\_url", document\_url: string, document\_name?: string} for a PDF, or \{type: "image\_url", image\_url: string} for an image); pages (array of 0-based page integers).                                                                                                     |

## Permissions

Every action above must be explicitly granted to an agent before it can be used. See [Permissions](/integrations/permissions) for the grant model and [Security](/integrations/security) for how credentials are protected.