Mistral AI API Guide 2026: Mixtral, Codestral, Embeddings

Mistral AI offers strong multilingual models at significantly lower cost.

Models

ModelContextBest ForCost/1M

Mistral Small32KSimple tasks$0.20/$0.60 Mistral Large 2128KComplex reasoning$3/$9 Mixtral 8x22B64KCost-efficient$2/$6 Codestral32KCode generation$1/$3

Basic Chat

python
from mistralai import Mistral
client = Mistral(api_key='your-api-key')
response = client.chat.complete(
    model='mistral-large-latest',
    messages=[
        {'role': 'system', 'content': 'You are a helpful assistant.'},
        {'role': 'user', 'content': 'Explain transformer attention'}
    ]
)
print(response.choices[0].message.content)
Streaming
for chunk in client.chat.stream(
    model='mistral-large-latest',
    messages=[{'role': 'user', 'content': 'Write a FastAPI tutorial'}]
):
    print(chunk.data.choices[0].delta.content, end='', flush=True)

Function Calling

python
import json
tools = [{'type': 'function', 'function': {
    'name': 'get_weather',
    'description': 'Get weather for a city',
    'parameters': {
        'type': 'object',
        'properties': {'city': {'type': 'string'}},
        'required': ['city']
    }
}}]
r = client.chat.complete(
    model='mistral-large-latest',
    messages=[{'role': 'user', 'content': 'Weather in Paris?'}],
    tools=tools, tool_choice='auto'
)
tc = r.choices[0].message.tool_calls[0]
print(json.loads(tc.function.arguments))

Embeddings for RAG

python
import numpy as np
def embed(texts):
    r = client.embeddings.create(model='mistral-embed', inputs=texts)
    return [item.embedding for item in r.data]
def cosine(a, b):
    return np.dot(a, b) / (np.linalg.norm(a) * np.linalg.norm(b))
docs = ['LangChain is a framework', 'Mistral builds open LLMs']
embs = embed(docs)
q_emb = embed(['What is Mistral?'])[0]
best = max(zip(docs, [cosine(q_emb, e) for e in embs]), key=lambda x: x[1])
print(best[0])  # Most relevant doc

Codestral

python
Best for 80+ programming languages
r = client.chat.complete(
    model='codestral-latest',
    messages=[{'role': 'user', 'content': 'Write async Python with aiohttp'}]
)
Fill-in-the-middle completion
r = client.fim.complete(
    model='codestral-latest',
    prompt='def sort_by_date(items):',
    suffix='    return sorted_items'
)

Local with Ollama

bash
ollama pull mistral:7b

python
from openai import OpenAI
c = OpenAI(base_url='http://localhost:11434/v1', api_key='ollama')
r = c.chat.completions.create(model='mistral:7b', messages=[{'role':'user','content':'Hello'}])

Conclusion

Mistral is the top choice for European data residency and cost-optimized multilingual AI. Codestral competes with GitHub Copilot at lower cost.