Mistral AI Documentation
All our documentation in your hands: build, customize, and deploy AI, your way.
Discover our platform
Three products, one platform. Pick where you want to start.
Vibe
The unified agent for productivity and coding. Chat with it on the web or mobile, or run it in your terminal and editor.
Studio
The developer console and the Mistral API. Keys, the Playground, evaluations, agents, and SDKs to ship applications.
Admin
The control plane for organization setup, billing, SSO, Workspaces, and access policies.
Latest models
Mistral Large 4
A state-of-the-art, open-weight, general-purpose multimodal model.
Z.ai GLM 5.3
A third-party open weight text model from Z.ai with a 1M-token context window.
Z.ai GLM 5.3 Flash
A third-party multimodal model from Z.ai with a 1M-token context window, optimized for agentic workflows and visual coding.
DeepSeek V4.1 Flash 1017
A third-party open weight model from DeepSeek with text and image inputs and a 1M-token context window, optimized for long-context coding and agentic workflows.
Mistral Small 4
Hybrid model unifying instruct, reasoning, and coding in a single efficient model.
Voxtral Mini Transcribe 3
Strong transcription and diarization at a competitive price point.
OCR 4.1
Our latest OCR service with paragraph-level bounding boxes, structural block labels, and block-level confidence scores.
Quickstarts
Self-contained walkthroughs grouped by product and role. Most take 15 minutes or less.
Run your first Vibe Work task
Open Vibe Work, run a multi-step task end-to-end, and review the result.
Analyze a dataset
Upload a spreadsheet and ask questions in plain language. Work writes and runs code.
Create your first Skill
Package a repeatable method into a Skill so Work applies the same procedure every time.
Install the Vibe CLI
Install the CLI, configure your API key, and send your first prompt from the terminal.
Scaffold a project with Vibe Code
Describe a project in natural language and let Vibe Code generate the files and structure.
Transcribe an audio file
Turn an audio file into accurate text with a single API call.
Transcribe speech in realtime
Stream live audio and get a transcript with low latency.
Generate speech
Convert text into natural speech and play it back with one API call.
Build a voice agent
Talk with an agent from your terminal, in a few steps.
Create your first API key
Create an API key in Studio, store it, and test it with one request.
Compare models in the playground
Send prompts, adjust parameters, and compare model outputs in the Studio playground.
Install the Mistral CLI
Install the CLI, sign in, and scaffold your first workflows project.
Build your first workflow
Scaffold a durable AI pipeline, run a worker, and trigger your first execution.
Build and deploy an agent end to end
Chain the Studio quickstarts: build an agent with the Agents SDK, add context, evaluate, deploy, and observe it.
Build your first app
Create a full-stack app, run it locally, and deploy it to Studio.
Create and use a prompt
Store a prompt in Studio, use it from your code, then promote a new version without redeploying.
Create a Skill in Studio
Package reusable instructions and files into a Skill for repeated tasks.
Add a connector
Register a connector, call one of its tools, and hand it to an agent.
Create a library and ask it a question
Add a document to a library and get an answer grounded in it.
Create and query a search index
Create a search index, ingest documents, and run your first query.
Build a RAG pipeline with Search Toolkit
Define a schema, ingest documents, and search them with the Search Toolkit.
Deploy a workflow to Mistral
Push a workflows project to GitHub and let Mistral build and run the worker.
Run your first evaluation
Evaluate a prompt on a small dataset, then read the scores in Studio.
Send your first trace
Instrument a Python script and find its trace in the Trace Explorer.
Score live traffic with online evaluations
Create a scorer and a live eval, send a span, and read its score.
Moderate an input
Classify inputs with the moderation API, then block unsafe ones with a guardrail.
Send your first API request
Install the SDK, set your API key, and get a model response.
Call your own functions
Let a model call a Python function of yours and answer with its result.
Get structured output
Extract a validated Python object from free text with a Pydantic schema.
Analyze an image
Send an image with a question and get a description.
Find the closest text with embeddings
Embed sentences and find the one closest in meaning to a question.
Run a batch job
Send a file of requests as one batch job and read the answers.
Create your organization
Set up billing, workspaces, and invite your first team members.
Configure SSO
Set up SAML Single Sign-On for enterprise authentication.
Manage workspaces & API keys
Create isolated workspaces, generate keys, and set usage limits.