Mistral AI

https://docs.mistral.ai/
Admin
Mistral AI provides a platform and a range of state-of-the-art open-source and commercial Large...
Tokens:25,645
Snippets:409
Trust Score:10
Update:3 months ago
Show doc for...
Tokens:
Raw
### Install Dependencies and Run Flask App

Source: https://docs.mistral.ai/getting-started/quickstarts/vibe-code/scaffold-a-project

Install the required Python packages and start the Flask development server.

```bash
pip install -r requirements.txt
python app.py
```

--------------------------------

### Install Vibe CLI Manually with uv

Source: https://docs.mistral.ai/getting-started/quickstarts/vibe-code/install-cli

Use this command to install the Vibe CLI if you prefer manual setup or are on Windows. Ensure Python 3.12+ is installed first.

```bash
# With uv (recommended)
uv tool install mistral-vibe
```

--------------------------------

### Install Vibe CLI with One-liner

Source: https://docs.mistral.ai/getting-started/quickstarts/vibe-code/install-cli

Run this command in your terminal for a quick installation on macOS or Linux. It handles checking for and installing dependencies like `uv`.

```bash
curl -LsSf https://mistral.ai/vibe/install.sh | bash
```

--------------------------------

### Launch Vibe CLI Setup Wizard

Source: https://docs.mistral.ai/getting-started/quickstarts/vibe-code/install-cli

Run this command in your project directory to launch the Vibe CLI. On the first run, it initiates a setup wizard to register your API key.

```bash
vibe
```

--------------------------------

### Re-run Vibe CLI Setup Wizard

Source: https://docs.mistral.ai/getting-started/quickstarts/vibe-code/install-cli

Use this command to re-initiate the setup wizard at any time. This is useful for re-registering API keys or changing configurations.

```bash
vibe --setup
```

--------------------------------

### Chat Completion Example

Source: https://docs.mistral.ai/api

Demonstrates how to perform a chat completion request. This example shows a basic setup for generating a response.

```python
from mistralai.client import MistralClient
from mistralai.models.chat_completion import ChatMessage

api_key = "YOUR_API_KEY"
model = "mistral-tiny"

client = MistralClient(api_key=api_key)

chat_response = client.chat(
    model=model,
    messages=[
        ChatMessage(role="user", content="Who are you?")
    ]
)

print(chat_response)
```

--------------------------------

### Create Project Directory and Launch Vibe CLI

Source: https://docs.mistral.ai/getting-started/quickstarts/vibe-code/scaffold-a-project

Start in a new directory and initialize the Vibe CLI to begin scaffolding a project from scratch.

```bash
mkdir my-project && cd my-project
vibe
```

--------------------------------

### Run Python Script

Source: https://docs.mistral.ai/getting-started/quickstarts/developer/first-api-request

Execute the Python quickstart script from your terminal.

```bash
python quickstart.py
```

--------------------------------

### Install Mistral AI TypeScript SDK

Source: https://docs.mistral.ai/getting-started/quickstarts/developer/first-api-request

Install the Mistral AI TypeScript SDK using npm.

```bash
npm install @mistralai/mistralai
```

--------------------------------

### Install Mistral AI Python SDK

Source: https://docs.mistral.ai/getting-started/quickstarts/developer/first-api-request

Install the Mistral AI Python SDK using pip.

```bash
pip install mistralai
```

--------------------------------

### Verify Vibe CLI Installation

Source: https://docs.mistral.ai/getting-started/quickstarts/vibe-code/install-cli

Check if the Vibe CLI is installed correctly by running this command. It should return a version number if the installation was successful.

```bash
vibe --version
```

--------------------------------

### Install Vibe CLI Manually with pip

Source: https://docs.mistral.ai/getting-started/quickstarts/vibe-code/install-cli

This command installs the Vibe CLI using pip. It's an alternative to using `uv` for manual installations, requiring Python 3.12+.

```bash
# With pip
pip install mistral-vibe
```

--------------------------------

### Run TypeScript Script

Source: https://docs.mistral.ai/getting-started/quickstarts/developer/first-api-request

Execute the TypeScript quickstart script using tsx.

```bash
npx tsx quickstart.ts
```

--------------------------------

### Python Chat Completion Example

Source: https://docs.mistral.ai/resources/sdks

Demonstrates how to perform a chat completion using the Python SDK. This example shows the structure of a typical API response.

```json
{
  "id": "7c163b08d98e4eebb35564ae14ffe6e4",
  "created": 1756130273,
  "model": "mistral-medium-latest",
  "usage": {
    "prompt_tokens": 10,
    "total_tokens": 1160,
    "completion_tokens": 1150
  },
  "object": "chat.completion",
  "choices": [
    {
      "index": 0,
      "finish_reason": "stop",
      "message": {
        "role": "assistant",
        "tool_calls": null,
        "content": "The \"best\" French cheese is highly subjective—it depends on personal taste, texture preferences, and how you plan to enjoy it (on a cheese board, melted, in cooking, etc.). However, here are some of the most **iconic, beloved, and award-winning French cheeses**, each celebrated for its unique qualities:\n\n### **Top Contenders for \"Best\" French Cheese**\n1. **Comté** (AOP)\n   - **Why?** A versatile, nutty, and complex **alpine cheese** aged 4 months to 3+ years. Younger Comté is mild and buttery; aged versions develop crystalline crunch and deep caramelized flavors.\n   - **Best for:** Cheese boards, fondue, grilled cheese, or eating plain.\n   - **Fun fact:** Made in the Jura Mountains, each wheel is unique based on the terroir and aging.\n\n2. **Roquefort** (AOP)\n   - **Why?** The **king of blue cheeses**, made from sheep’s milk and aged in natural caves. Creamy, tangy, and intensely flavorful with a sharp, salty bite.\n   - **Best for:** Spread on bread with honey or pears, in salads, or sauces.\n   - **Fun fact:** One of the oldest cheeses in France (mentioned by Pliny the Elder in 79 AD).\n\n3. **Camembert de Normandie** (AOP)\n   - **Why?** The **quintessential French cheese**—creamy, earthy, and slightly mushroomy when ripe. Invented in 1791 by Marie Harel.\n   - **Best for:** Baked (Camembert en croûte), with crusty bread, or apples.\n   - **Pro tip:** Eat it at room temperature when the center is soft but not runny.\n\n4. **Époisses de Bourgogne** (AOP)\n   - **Why?** A **pungent, washed-rind cheese** with a velvety texture and bold, barnyardy flavor. Napoleon reportedly loved it.\n   - **Best for:** Spreading on bread or pairing with Burgundy wine.\n   - **Warning:** So strong it’s banned on French public transport!\n\n5. **Reblochon** (AOP)\n   - **Why?** The **star of tartiflette**—creamy, nutty, and slightly smoky. Made in the Alps with raw cow’s milk.\n   - **Best for:** Melting in tartiflette, raclette, or on potatoes.\n\n6. **Brie de Meaux** (AOP)\n   - **Why?** The **\"King of Cheeses\"** (as declared at the 1815 Congress of Vienna). Rich, buttery, and mushroomy with a soft rind.\n   - **Best for:** Pairing with Champagne or fruit (grapes, figs).\n\n7. **Munster (Munster-Géromé)** (AOP)\n   - **Why?** A **stinky, washed-rind cheese** from Alsace with a creamy, tangy interior. Often served with cumin.\n   - **Best for:** Melting in tarts or eating with beer.\n\n8. **Beaufort** (AOP)\n   - **Why?** A **firm, nutty alpine cheese** similar to Gruyère but richer. Aged versions are caramelized and complex.\n   - **Best for:** Fondue, gratins, or shaving over salads.\n\n9. **Pélardon** (AOP)\n   - **Why?** A **small, tangy goat cheese** from the Cévennes or Provence. Comes in fresh (soft) or aged (firm, peppery) varieties.\n   - **Best for:** Salads, tartines, or grilled with herbs.\n\n10. **Bleu d’Auvergne** (AOP)\n   - **Why?** A **creamy, spicy blue cheese** with a sharper taste than Roquefort but less crumbly.\n   - **Best for:** Dressings, burgers, or cheese plates with walnuts.\n\n### **How to Choose?**\n- **Mild & creamy?** → Brie de Meaux, Camembert, Saint-André.\n- **Bold & stinky?** → Époisses, Munster, Pont-l’Évêque.\n- **Nutty & firm?** → Comté, Beaufort, Cantal.\n- **Blue cheese lover?** → Roquefort, Bleu d’Auvergne.\n- **Goat cheese fan?** → Crottin de Chavignol, Pélardon, Valençay.\n\n### **Pro Tips for Enjoying French Cheese**\n- **Serve at room temperature** (take it out of the fridge 1–2 hours before eating).\n- **Pair with wine**: Brie + Champagne, Roquefort + Sauternes, Comté + Jura wine.\n- **Avoid plastic wrap**—store in cheese paper or a breathable container.\n- **Visit a *fromagerie*** in France for expert recommendations!\n\n### **Final Verdict**\nIf forced to pick **one \"best\"**, **Comté** (aged 24+ months) and **Roquefort** are often crowned for their depth and versatility. But the true joy of French cheese is in the **diversity**—so try them all!\n\n**What’s your favorite?** 😊🧀"
      }
    }
  ]
}
```

--------------------------------

### Start the Mistral AI Workflow Worker

Source: https://docs.mistral.ai/getting-started/quickstarts/developer/build-a-workflow

Run this command from your project's root directory to start the workflow worker. The worker connects to the Mistral API, registers your defined workflows, and waits for execution tasks. Keep this terminal running while your workflows are active.

```bash
make start-worker
```

--------------------------------

### Create a Library and upload a document (Python)

Source: https://docs.mistral.ai/getting-started/quickstarts/developer/rag-document-search

Use this snippet to create a new library and upload a document for retrieval. Ensure you have your Mistral API key set as an environment variable and the Mistral SDK installed.

```python
import os
from mistralai import Mistral

client = Mistral(api_key=os.environ["MISTRAL_API_KEY"])

# Create a library
library = client.files.upload(
    file=open("company-handbook.pdf", "rb"),
    purpose="retrieval",
)
print(f"File uploaded: {library.id}")
```

--------------------------------

### Define a Minimal Hello World Workflow

Source: https://docs.mistral.ai/getting-started/quickstarts/developer/build-a-workflow

This Python code defines a simple workflow with an activity and an entrypoint. The `greet` activity performs a string operation, and the `HelloWorkflow` orchestrates its execution. Use this as a starting point for more complex workflows.

```python
from pydantic import BaseModel
import mistralai.workflows as workflows

class HelloInput(BaseModel):
    name: str = "World"

@workflows.activity()
async def greet(name: str) -> str:
    return f"Hello, {name}! Welcome to Mistral Workflows."

@workflows.workflow.define(
    name="hello-world",
    workflow_display_name="Hello World",
    workflow_description="A minimal hello-world workflow.",
)
class HelloWorkflow:
    @workflows.workflow.entrypoint
    async def run(self, input: HelloInput) -> str:
        return await greet(input.name)
```

--------------------------------

### Chat Completion Example in Python

Source: https://docs.mistral.ai/api

Shows how to make a chat completion request using the Mistral AI Python SDK. The API key should be provided via the MISTRAL_API_KEY environment variable.

```python
import os
from mistralai.client import Mistral

client = Mistral(api_key=os.environ["MISTRAL_API_KEY"])

response = client.chat.complete(
    model="mistral-large-latest",
    messages=[
        {
            "role": "user",
            "content": "Who is the best French painter? Answer in one short sentence.",
        }
    ],
)

print(response.choices[0].message.content)

```

--------------------------------

### Create a Library and upload a document (TypeScript)

Source: https://docs.mistral.ai/getting-started/quickstarts/developer/rag-document-search

Use this snippet to create a new library and upload a document for retrieval. Ensure you have your Mistral API key set as an environment variable and the Mistral SDK installed.

```typescript
const client = new Mistral({ apiKey: process.env.MISTRAL_API_KEY });

const library = await client.files.upload({
  file: fs.createReadStream('company-handbook.pdf'),
  purpose: 'retrieval',
});
console.log(`File uploaded: ${library.id}`);
```

--------------------------------

### Chat Completion Example in TypeScript

Source: https://docs.mistral.ai/api

Demonstrates how to perform a chat completion request using the Mistral AI SDK in TypeScript. Ensure your MISTRAL_API_KEY is set in the environment variables.

```typescript
import { Mistral } from "@mistralai/mistralai";

const client = new Mistral({ apiKey: process.env.MISTRAL_API_KEY });

const response = await client.chat.complete({
  model: "mistral-large-latest",
  messages: [
    {
      role: "user",
      content: "Who is the best French painter? Answer in one short sentence.",
    },
  ],
});

console.log(response.choices[0].message.content);

```

--------------------------------

### Python Chat Completion (V1)

Source: https://docs.mistral.ai/resources/sdks

Example of performing a chat completion using the Mistral AI Python SDK (V1). Ensure the MISTRAL_API_KEY environment variable is set.

```python
import os
from mistralai import Mistral

api_key = os.environ["MISTRAL_API_KEY"]
model = "mistral-medium-latest"

client = Mistral(api_key=api_key)

chat_response = client.chat.complete(
    model = model,
    messages = [
        {
            "role": "user",
            "content": "What is the best French cheese?",
        },
    ]
)
```

--------------------------------

### Chat Completions

Source: https://docs.mistral.ai/api

This section details how to use the chat completions endpoint, including available parameters and example requests in various languages.

```APIDOC
## POST /v1/chat/completions

### Description
This endpoint allows you to generate chat completions using Mistral AI models. You can specify various parameters to control the generation process, such as the model, messages, temperature, and tools.

### Method
POST

### Endpoint
https://api.mistral.ai/v1/chat/completions

### Parameters
#### Request Body
- **model** (string) - Required - The model to use for chat completions.
- **messages** (array) - Required - A list of messages comprising the conversation.
  - **role** (string) - Required - The role of the author of a message (`user`, `assistant`, or `system`).
  - **content** (string) - Required - The content of the message.
- **temperature** (number|null) - Optional - Sampling temperature. Recommended between 0.0 and 0.7. Higher values make output more random, lower values make it more focused.
- **top_p** (number|null) - Optional - Nucleus sampling. Considers tokens with `top_p` probability mass. We recommend altering this or `temperature` but not both.
- **tools** (array<Tool|WebSearchTool|WebSearchPremiumTool|CodeInterpreterTool|ImageGenerationTool|DocumentLibraryTool|CustomConnector>|null) - Optional - A list of tools the model may call.
- **tool_choice** (ToolChoice|"auto"|"none"|"any"|"required") - Optional - Controls which (if any) tool is called by the model. `none` means the model will not call any tool. `auto` means the model can pick between generating a message or calling one or more tools. `any` or `required` means the model must call one or more tools. Specifying a particular tool forces the model to call that tool.

### Request Example
```json
{
  "model": "mistral-large-latest",
  "messages": [
    {
      "role": "user",
      "content": "Who is the best French painter? Answer in one short sentence."
    }
  ]
}
```

### Response
#### Success Response (200 application/json)
- **id** (string) - Unique identifier for the completion.
- **object** (string) - Type of object returned, e.g., `chat.completion`.
- **created** (integer) - Timestamp of when the completion was created.
- **model** (string) - The model used for the completion.
- **choices** (array<ChatCompletionChoice>) - A list of completion choices.
- **usage** (UsageInfo) - Information about the token usage.

#### Success Response (200 text/event-stream)
- **event-stream** (CompletionEvent) - A stream of completion events.

#### Response Example (application/json)
```json
{
  "choices": [
    {
      "finish_reason": "stop",
      "index": 0,
      "message": {
        "content": "Claude Monet is widely considered one of the best French painters, known for his pioneering role in Impressionism.",
        "role": "assistant"
      }
    }
  ],
  "created": 1702256327,
  "id": "cmpl-e5cc70bb28c444948073e77776eb30ef",
  "model": "mistral-small-latest",
  "object": "chat.completion",
  "usage": {
    "prompt_tokens": 25,
    "completion_tokens": 30,
    "total_tokens": 55
  }
}
```
```

```typescript
import { Mistral } from "@mistralai/mistralai";

const client = new Mistral({ apiKey: process.env.MISTRAL_API_KEY });

const response = await client.chat.complete({
  model: "mistral-large-latest",
  messages: [
    {
      role: "user",
      content: "Who is the best French painter? Answer in one short sentence.",
    },
  ],
});

console.log(response.choices[0].message.content);
```

```python
import os
from mistralai.client import Mistral

client = Mistral(api_key=os.environ["MISTRAL_API_KEY"])

response = client.chat.complete(
    model="mistral-large-latest",
    messages=[
        {
            "role": "user",
            "content": "Who is the best French painter? Answer in one short sentence.",
        }
    ],
)

print(response.choices[0].message.content)
```

```curl
curl https://api.mistral.ai/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $MISTRAL_API_KEY" \
  -d '{ \
    "model": "mistral-large-latest", \
    "messages": [ \
      { \
        "role": "user", \
        "content": "Who is the best French painter? Answer in one short sentence." \
      } \
    ] \
  }'
```

--------------------------------

### Python Chat Completion (V2)

Source: https://docs.mistral.ai/resources/sdks

Example of performing a chat completion using the Mistral AI Python SDK (V2). Ensure the MISTRAL_API_KEY environment variable is set.

```python
import os
from mistralai.client import Mistral

api_key = os.environ["MISTRAL_API_KEY"]
model = "mistral-medium-latest"

client = Mistral(api_key=api_key)

chat_response = client.chat.complete(
    model = model,
    messages = [
        {
            "role": "user",
            "content": "What is the best French cheese?",
        },
    ]
)
```

--------------------------------

### Chat Completion with Tool Calls

Source: https://docs.mistral.ai/api

Example of enabling parallel tool calls in chat completions. The model can invoke multiple tools simultaneously.

```python
from mistralai.client import MistralClient
from mistralai.models.chat_completion import ChatMessage

api_key = "YOUR_API_KEY"
model = "mistral-tiny"

client = MistralClient(api_key=api_key)

chat_response = client.chat(
    model=model,
    messages=[
        ChatMessage(role="user", content="What is the weather in Paris and what is the capital of France?")
    ],
    parallel_tool_calls=True
)

print(chat_response)
```

--------------------------------

### Define a get_weather tool schema in Python

Source: https://docs.mistral.ai/getting-started/quickstarts/developer/build-an-agent

Define the schema for a tool that the model can call. This Python example creates a 'get_weather' tool that accepts a city name as a parameter.

```python
import json
import os
from mistralai import Mistral

client = Mistral(api_key=os.environ["MISTRAL_API_KEY"])

# Define the tool schema
tools = [
    {
        "type": "function",
        "function": {
            "name": "get_weather",
            "description": "Get the current weather for a given city.",
            "parameters": {
                "type": "object",
                "properties": {
                    "city": {
                        "type": "string",
                        "description": "The city name, e.g. 'Paris'."
                    }
                },
                "required": ["city"]
            }
        }
    }
]
```

--------------------------------

### Create a Library and upload a document (curl)

Source: https://docs.mistral.ai/getting-started/quickstarts/developer/rag-document-search

Use this command to create a new library and upload a document for retrieval. Ensure your Mistral API key is set as an environment variable.

```bash
curl https://api.mistral.ai/v1/files \
  -H "Authorization: Bearer $MISTRAL_API_KEY" \
  -F "file=@company-handbook.pdf" \
  -F "purpose=retrieval"
```

--------------------------------

### Define a get_weather tool schema in TypeScript

Source: https://docs.mistral.ai/getting-started/quickstarts/developer/build-an-agent

Define the schema for a tool that the model can call. This TypeScript example creates a 'get_weather' tool that accepts a city name as a parameter.

```typescript
const client = new Mistral({ apiKey: process.env.MISTRAL_API_KEY });

const tools = [
  {
    type: 'function' as const,
    function: {
      name: 'get_weather',
      description: 'Get the current weather for a given city.',
      parameters: {
        type: 'object',
        properties: {
          city: {
            type: 'string',
            description: "The city name, e.g. 'Paris'.",
          },
        },
        required: ['city'],
      },
    },
  },
];
```

--------------------------------

### Initialize Mistral Client and Chat Completion

Source: https://docs.mistral.ai/resources/sdks

This snippet shows how to initialize the Mistral client using an API key and make a chat completion request. Ensure the MISTRAL_API_KEY environment variable is set or replace 'your_api_key' with your actual key.

```typescript
const apiKey = process.env.MISTRAL_API_KEY || 'your_api_key';

const client = new Mistral({apiKey: apiKey});

const chatResponse = await client.chat.complete({
  model: 'mistral-medium-latest',
  messages: [{role: 'user', content: 'What is the best French cheese?'}],
});
```

```json
{
  "id": "7c163b08d98e4eebb35564ae14ffe6e4",
  "created": 1756130273,
  "model": "mistral-medium-latest",
  "usage": {
    "prompt_tokens": 10,
    "total_tokens": 1160,
    "completion_tokens": 1150
  },
  "object": "chat.completion",
  "choices": [
    {
      "index": 0,
      "finish_reason": "stop",
      "message": {
        "role": "assistant",
        "tool_calls": null,
        "content": "The \"best\" French cheese is highly subjective—it depends on personal taste, texture preferences, and how you plan to enjoy it (on a cheese board, melted, in cooking, etc.). However, here are some of the most **iconic, beloved, and award-winning French cheeses**, each celebrated for its unique qualities:\n\n### **Top Contenders for \"Best\" French Cheese**\n1. **Comté** (AOP)\n   - **Why?** A versatile, nutty, and complex **alpine cheese** aged 4 months to 3+ years. Younger Comté is mild and buttery; aged versions develop crystalline crunch and deep caramelized flavors.\n   - **Best for:** Cheese boards, fondue, grilled cheese, or eating plain.\n   - **Fun fact:** Made in the Jura Mountains, each wheel is unique based on the terroir and aging.\n\n2. **Roquefort** (AOP)\n   - **Why?** The **king of blue cheeses**, made from sheep’s milk and aged in natural caves. Creamy, tangy, and intensely flavorful with a sharp, salty bite.\n   - **Best for:** Spread on bread with honey or pears, in salads, or sauces.\n   - **Fun fact:** One of the oldest cheeses in France (mentioned by Pliny the Elder in 79 AD).\n\n3. **Camembert de Normandie** (AOP)\n   - **Why?** The **quintessential French cheese**—creamy, earthy, and slightly mushroomy when ripe. Invented in 1791 by Marie Harel.\n   - **Best for:** Baked (Camembert en croûte), with crusty bread, or apples.\n   - **Pro tip:** Eat it at room temperature when the center is soft but not runny.\n\n4. **Époisses de Bourgogne** (AOP)\n   - **Why?** A **pungent, washed-rind cheese** with a velvety texture and bold, barnyardy flavor. Napoleon reportedly loved it.\n   - **Best for:** Spreading on bread or pairing with Burgundy wine.\n   - **Warning:** So strong it’s banned on French public transport!\n\n5. **Reblochon** (AOP)\n   - **Why?** The **star of tartiflette**—creamy, nutty, and slightly smoky. Made in the Alps with raw cow’s milk.\n   - **Best for:** Melting in tartiflette, raclette, or on potatoes.\n\n6. **Brie de Meaux** (AOP)\n   - **Why?** The **\"King of Cheeses\"** (as declared at the 1815 Congress of Vienna). Rich, buttery, and mushroomy with a soft rind.\n   - **Best for:** Pairing with Champagne or fruit (grapes, figs).\n\n7. **Munster (Munster-Géromé)** (AOP)\n   - **Why?** A **stinky, washed-rind cheese** from Alsace with a creamy, tangy interior. Often served with cumin.\n   - **Best for:** Melting in tarts or eating with beer.\n\n8. **Beaufort** (AOP)\n   - **Why?** A **firm, nutty alpine cheese** similar to Gruyère but richer. Aged versions are caramelized and complex.\n   - **Best for:** Fondue, gratins, or shaving over salads.\n\n9. **Pélardon** (AOP)\n   - **Why?** A **small, tangy goat cheese** from the Cévennes or Provence. Comes in fresh (soft) or aged (firm, peppery) varieties.\n   - **Best for:** Salads, tartines, or grilled with herbs.\n\n10. **Bleu d’Auvergne** (AOP)\n   - **Why?** A **creamy, spicy blue cheese** with a sharper taste than Roquefort but less crumbly.\n   - **Best for:** Dressings, burgers, or cheese plates with walnuts.\n\n### **How to Choose?**\n- **Mild & creamy?** → Brie de Meaux, Camembert, Saint-André.\n- **Bold & stinky?** → Époisses, Munster, Pont-l’Évêque.\n- **Nutty & firm?** → Comté, Beaufort, Cantal.\n- **Blue cheese lover?** → Roquefort, Bleu d’Auvergne.\n- **Goat cheese fan?** → Crottin de Chavignol, Pélardon, Valençay.\n\n### **Pro Tips for Enjoying French Cheese**\n- **Serve at room temperature** (take it out of the fridge 1–2 hours before eating).\n- **Pair with wine**: Brie + Champagne, Roquefort + Sauternes, Comté + Jura wine.\n- **Avoid plastic wrap**—store in cheese paper or a breathable container.\n- **Visit a *fromagerie*** in France for expert recommendations!\n\n### **Final Verdict**\nIf forced to pick **one \"best\"**, **Comté** (aged 24+ months) and **Roquefort** are often crowned for their depth and versatility. But the true joy of French cheese is in the **diversity**—so try them all!\n\n**What’s your favorite?** 😊🧀"
      }
    }
  ]
}
```

--------------------------------

### Execute function and return result to model in Python

Source: https://docs.mistral.ai/getting-started/quickstarts/developer/build-an-agent

Simulate executing the tool function with the arguments provided by the model, then send the result back to the model to generate a final answer.

```python
# Simulate the function (replace with a real API call)
def get_weather(city: str) -> dict:
    return {"city": city, "temperature": "18°C", "condition": "Partly cloudy"}

# Execute the tool call
args = json.loads(tool_call.function.arguments)
result = get_weather(**args)

# Send the result back to the model
messages.append(response.choices[0].message)
messages.append({
    "role": "tool",
    "name": tool_call.function.name,
    "content": json.dumps(result),
    "tool_call_id": tool_call.id,
})

final_response = client.chat.complete(
    model="mistral-medium-latest",
    messages=messages,
    tools=tools,
)

print(final_response.choices[0].message.content)
# "The weather in Paris is 18°C and partly cloudy."
```

--------------------------------

### Add DNS TXT Record for Domain Verification

Source: https://docs.mistral.ai/getting-started/quickstarts/admin/configure-sso

Add this TXT record to your domain's DNS configuration to prove ownership. DNS propagation can take up to 48 hours.

```text
Type:  TXT
Name:  _mistral-verification
Value: mistral-verify=abc123xyz...
```

--------------------------------

### Successful JSON Response Structure

Source: https://docs.mistral.ai/api

Example of a successful JSON response from the chat completions endpoint. This structure includes details about the choices, creation time, ID, model used, object type, and usage statistics.

```json
{
  "choices": [
    {
      "finish_reason": "stop",
      "index": "<to fill>"
    }
  ],
  "created": "1702256327",
  "id": "cmpl-e5cc70bb28c444948073e77776eb30ef",
  "model": "mistral-small-latest",
  "object": "chat.completion",
  "usage": {}
}

```

--------------------------------

### Scaffold a Mistral AI Workflow Project

Source: https://docs.mistral.ai/getting-started/quickstarts/developer/build-a-workflow

Use the Mistral AI CLI to set up a new Python project for building workflows. This command initializes the project structure and prompts for your Mistral API key.

```bash
uvx mistralai-workflows-cli setup
```

--------------------------------

### Chat Completion with Custom Stop Sequences

Source: https://docs.mistral.ai/api

Shows how to use custom stop sequences to halt generation. This can be a single string or an array of strings.

```python
from mistralai.client import MistralClient
from mistralai.models.chat_completion import ChatMessage

api_key = "YOUR_API_KEY"
model = "mistral-tiny"

client = MistralClient(api_key=api_key)

chat_response = client.chat(
    model=model,
    messages=[
        ChatMessage(role="user", content="Write a short poem about the sea.")
    ],
    stop=[".", "!", "?"]
)

print(chat_response)
```

--------------------------------

### Build an Agent with Tools

Source: https://docs.mistral.ai/getting-started/quickstarts/developer/build-an-agent

This TypeScript code simulates a weather function, executes a tool call, and sends the result back to the model. Ensure you have the necessary client and message structures set up.

```typescript
// Simulate the function (replace with a real API call)
function getWeather(city: string) {
  return { city, temperature: '18°C', condition: 'Partly cloudy' };
}

// Execute the tool call
const args = JSON.parse(toolCall.function.arguments);
const result = getWeather(args.city);

// Send the result back to the model
messages.push(response.choices[0].message);
messages.push({
  role: 'tool' as const,
  name: toolCall.function.name,
  content: JSON.stringify(result),
  toolCallId: toolCall.id,
});

const finalResponse = await client.chat.complete({
  model: 'mistral-medium-latest',
  messages,
  tools,
});

console.log(finalResponse.choices[0].message.content);
// "The weather in Paris is 18°C and partly cloudy."
```

--------------------------------

### Chat Completion with Streaming

Source: https://docs.mistral.ai/api

Illustrates how to use the streaming option for chat completions. This is useful for real-time responses.

```python
from mistralai.client import MistralClient
from mistralai.models.chat_completion import ChatMessage

api_key = "YOUR_API_KEY"
model = "mistral-tiny"

client = MistralClient(api_key=api_key)

small_chat_response = client.chat(
    model=model,
    messages=[
        ChatMessage(role="user", content="Explain the importance of low-latency LLMs.")
    ],
    stream=True
)

for chunk in small_chat_response:
    print(chunk)
```

--------------------------------

### Verify Domain Ownership with dig

Source: https://docs.mistral.ai/getting-started/quickstarts/admin/configure-sso

You can confirm the TXT record is live before clicking Verify by running this command in your terminal.

```bash
dig TXT _mistral-verification.acme.com
```

--------------------------------

### Chat Completion with JSON Mode

Source: https://docs.mistral.ai/api

Shows how to enable JSON mode for chat completions, ensuring the output is a valid JSON object. Requires instructing the model to output JSON.

```python
from mistralai.client import MistralClient
from mistralai.models.chat_completion import ChatMessage, ResponseFormat

api_key = "YOUR_API_KEY"
model = "mistral-tiny"

client = MistralClient(api_key=api_key)

chat_response = client.chat(
    model=model,
    response_format=ResponseFormat(type="json_object"),
    messages=[
        ChatMessage(role="user", content="Provide a JSON object with a list of fruits and their colors.")
    ]
)

print(chat_response)
```
Tokens:
Raw