Mistral Hello World

Mistral AI Hello World

Overview

Minimal working examples demonstrating Mistral AI chat completions, streaming, multi-turn conversation, and JSON mode. Uses the official @mistralai/mistralai TypeScript SDK and mistralai Python SDK.

Prerequisites

• Completed mistral-install-auth setup
• Valid MISTRAL_API_KEY environment variable set
• Node.js 18+ or Python 3.9+

Instructions

Step 1: Basic Chat Completion

TypeScript (hello-mistral.ts)

import { Mistral } from '@mistralai/mistralai';

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

async function main() {
  const response = await client.chat.complete({
    model: 'mistral-small-latest',
    messages: [
      { role: 'user', content: 'Say "Hello, World!" in a creative way.' },
    ],
  });

  console.log(response.choices?.[0]?.message?.content);
  console.log('Tokens used:', response.usage);
}

main().catch(console.error);

Python (hello_mistral.py)

import os
from mistralai import Mistral

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

response = client.chat.complete(
    model="mistral-small-latest",
    messages=[
        {"role": "user", "content": "Say 'Hello, World!' in a creative way."}
    ],
)

print(response.choices[0].message.content)
print(f"Tokens: {response.usage}")

Step 2: Run the Example

# TypeScript
npx tsx hello-mistral.ts

# Python
python hello_mistral.py

Step 3: Streaming Response

Streaming delivers the first token in ~200ms instead of waiting 1-2s for the full response.

TypeScript

import { Mistral } from '@mistralai/mistralai';

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

async function streamChat() {
  const stream = await client.chat.stream({
    model: 'mistral-small-latest',
    messages: [
      { role: 'user', content: 'Tell me a short story about AI.' },
    ],
  });

  for await (const event of stream) {
    const content = event.data?.choices?.[0]?.delta?.content;
    if (content) process.stdout.write(content);
  }
  console.log(); // newline
}

streamChat().catch(console.error);

Python

stream = client.chat.stream(
    model="mistral-small-latest",
    messages=[{"role": "user", "content": "Tell me a short story about AI."}],
)

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

Step 4: Multi-Turn Conversation

const messages: Array<{ role: 'system' | 'user' | 'assistant'; content: string }> = [
  { role: 'system', content: 'You are a helpful coding assistant.' },
  { role: 'user', content: 'What is the capital of France?' },
];

const r1 = await client.chat.complete({
  model: 'mistral-small-latest', messages,
});
const answer = r1.choices?.[0]?.message?.content ?? '';
console.log('A1:', answer);

// Continue the conversation
messages.push({ role: 'assistant', content: answer });
messages.push({ role: 'user', content: 'What about Germany?' });

const r2 = await client.chat.complete({
  model: 'mistral-small-latest', messages,
});
console.log('A2:', r2.choices?.[0]?.message?.content);

Step 5: JSON Mode (Structured Output)

const response = await client.chat.complete({
  model: 'mistral-small-latest',
  messages: [
    { role: 'user', content: 'List 3 programming languages with their year of creation as JSON.' },
  ],
  responseFormat: { type: 'json_object' },
});

const data = JSON.parse(response.choices?.[0]?.message?.content ?? '{}');
console.log(data);

Step 6: With Temperature and Token Limits

const response = await client.chat.complete({
  model: 'mistral-small-latest',
  messages: [{ role: 'user', content: 'Write a haiku about coding.' }],
  temperature: 0.7,   // 0-1, higher = more creative
  maxTokens: 100,     // cap output length
  topP: 0.9,          // nucleus sampling
});

Output

• Working code file with Mistral client initialization
• Successful API response with generated text
• Console output showing response and token usage

Error Handling

Model Quick Reference

Examples

Verify a new local API key

Set MISTRAL_API_KEY in a local, uncommitted environment file, run the first chat-completion example with a low token cap, and confirm the response is non-empty. If it returns 401, stop and correct the secret rather than retrying the request with a hard-coded key.

Resources

• Mistral AI Quickstart
• Chat Completions API
• Models Overview

Next Steps

Proceed to mistral-core-workflow-a for production chat patterns or mistral-local-dev-loop for dev workflow setup.