Mistral AI API
by Mistral AI
French AI models for text, documents and code.
1. Price
No free tier on the pricing page. As reported on the official pricing page, checked Oct 22, 2026.
mistral.ai/pricing2. Commercial use
“Mistral AI hereby assigns to Customer all right, title, and interest, if any, in and to Output that Mistral AI may have.”
Allowed on a paid account under the commercial terms; you own the output. Data: opt-out; API data can improve Mistral's services unless an admin turns the toggle off.
legal.mistral.ai/terms/commercial-terms-of-service/3. Key & sign-up
“Browser clients cannot store long-lived API keys safely and cannot set Authorization headers on WebSocket connections.”
API key sent as Authorization: Bearer
docs.mistral.ai/studio/audio/deprecated/speech_to_text/realtime_transcription/client_auth.md4. Agent access
“It exposes your workspace as MCP tools and authenticates each request with your Mistral API key.”
Our agent check: 4 of 6 (Nov 13, 2026).
docs.mistral.ai/resources/mcp.mdChecked against the maker's official pages, pricing on Oct 22, 2026. Tap a verdict to see its source.How we checkedReport an error
What Mistral AI API does
The Mistral AI API lets your app send text, images or documents to Mistral's AI models and get answers back. It covers chat, reading scanned documents, turning text into searchable numbers, speech and code. You pay per amount of text processed, after adding a payment method; the free plan does not include the API.
- Mistral AI API is one of 41 APIs in the Model hosting and inference category on AgentGid.
- It is not among the 13 that can be started for free.
- 6 of them ship an official MCP server; Mistral AI API does not.
- Its official SDK is the 3rd most downloaded in the category (10.2M a week).
What people build with it
Answers and summaries in your app
Add a chat box, a "summarise this" button or automatic replies, using a small model that costs little per message.
Turn emails into tidy records
Pull the customer, topic and priority out of a support email straight into the fields of your helpdesk.
Read scanned documents and PDFs
Turn invoices, contracts or scanned pages into text and tables your app can search and store.
Search help articles by meaning
Find the help article closest to a question, even when it uses other words, such as "money back" for a refund policy.
Process a large pile overnight
Send thousands of texts as one batch job when you do not need instant answers, at half the price.
Switch from OpenAI with small changes Idea
Move an existing chatbot to Mistral by changing the client, the setup line and the model name.
Mistral AI API popularity
SDK downloads from npm and PyPI, weekly. Numbers as of Nov 17, 2026.
Pricing
None
No monthly plan
e.g. Mistral Large $0.5 per million input tokens, $1.5 per million output; batch processing 50% off
What one action costs
| Action | Price |
|---|---|
| 1 million input tokens, Mistral Small 4 | $0.15 |
| 1 million output tokens, Mistral Small 4 | $0.6 |
| 1 million input tokens, Mistral Large 4 | $0.7 |
| 1,000 pages read with OCR 4.1 | $4 |
| 1 million tokens turned into embeddings (Mistral Embed) | $0.1 |
Price history
- A year of the pricing page in the Internet Archive (7 months): the lowest monthly figure went from $14.99 (2025-10) to $5.99 (2026-11).
- 2026-06: lowest $/month figure $14.99 → $5.99; new figures $5.99/mo
- 2026-07: no longer shown $50/mo
- 2026-08: new figures $50/mo
- 2026-09: new figures $10/mo, $15/mo, $30/mo
As reported on the official pricing page, checked Oct 22, 2026. Prices change: confirm before you buy. mistral.ai/pricing
Limits
Tokens per minute and requests per second per model, shown in the Admin Panel; higher limits through support
Monthly spending limit per Organization and per Workspace
From the official docs. docs.mistral.ai/admin/billing-usage/usage-limits.md
Key and access
API key sent as Authorization: Bearer
MISTRAL_API_KEYYes
Steps only a person can do
- Create a Mistral account and activate billing with a payment method and prepaid credit. Nothing in code; the API works only after this.
- Create an API key in Studio with an expiration date and copy it into .env. Read MISTRAL_API_KEY on the server; the SDK reads that variable name.
- Decide on training use; an admin can switch off API data training in the Admin Panel Privacy menu. Note the choice in the project README.
- Set a monthly spending limit for the Organization or Workspace. Handle 402 errors by telling the person billing needs attention.
Commercial use and your data
You pay per use, and you own what the models produce for you. Several models are also released as open weights, many under Apache 2.0, so you can run them yourself.
Allowed on a paid account under the commercial terms; you own the output.
“Mistral AI hereby assigns to Customer all right, title, and interest, if any, in and to Output that Mistral AI may have.” source
Text output is yours; image output may not be used to train an image generator that competes with Mistral.
“Customer may not use image Outputs to develop or train any image generation product that competes with a Mistral AI product.” source
Only open-weight models can ship with your app, under each model's license (Apache 2.0 for Mistral Small 4, the Mistral open-weight license for Large 4); API models stay online.
“Mistral Large 4 mistral-large-4 Open, Generalist” source
opt-out; API data can improve Mistral's services unless an admin turns the toggle off.
“Customers retain full control over this processing and have the right to opt out at any time.” source
Zero data retention is available on paid plans for supported stateless calls, on request; Mistral may approve or deny it.
“ZDR is available on paid plans for supported stateless API calls.” source
Errors to handle
The API key is missing, invalid, or expired
Check the key in Studio and the header format Authorization Bearer.
Billing blocks the request (no credit left, spending limit reached, failed payment)
Ask the person to top up credit or raise the spending limit.
Valid JSON but a wrong parameter value
Read the field-level errors in the response body and fix the request.
Rate limit for your tier exceeded
Retry with exponential backoff and respect the Retry-After header; use batch jobs for large volumes.
Service overloaded or under maintenance
Retry with exponential backoff and check status.mistral.ai.
Use it from an agent
4 of 6: llms.txt yes, OpenAPI yes, SDK yes, MCP no, status page no, AI assistants allowed yes
Checked Nov 13, 2026.
Mistral MCP server (hosted)
claude mcp add --transport http mistral https://api.mistral.ai/mcp \ --header 'Authorization: Bearer ${MISTRAL_API_KEY}'
npm install @mistralai/mistralaipip install mistralaiAlternatives
Mistral AI API This page
Pick the OpenAI API for GPT models with text, vision, tool calling and agents, priced per 1M tokens.
Pick the Gemini API when you want a free tier and text, image, audio and video input.
Pick Groq for fast inference of open models such as GPT-OSS and Llama, priced per 1M tokens.
Get started
Tell your AI assistant
“Add Mistral to my app on the server: install mistralai (V2+, import from mistralai.client), read MISTRAL_API_KEY from .env, use mistral-small-latest, retry 429 with backoff, and never send the key to the browser.”
Create a Mistral account, activate billing with a payment method, then create an API key in Studio; the key is shown only once.
From docs.mistral.ai/getting-started/quickstarts/studio/activate-and-generate-api-key.md
Store the key in the MISTRAL_API_KEY variable and install the official package.
export MISTRAL_API_KEY="your-api-key" uv add mistralaiFrom docs.mistral.ai/getting-started/quickstarts/studio/activate-and-generate-api-key.md
Send a first chat request (the Python example is in the first use above).
curl https://api.mistral.ai/v1/chat/completions \ -H "Authorization: Bearer $MISTRAL_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "mistral-small-latest", "messages": [{"role": "user", "content": "Hello, Mistral!"}] }'From docs.mistral.ai/getting-started/quickstarts/studio/activate-and-generate-api-key.md
First request
curl https://api.mistral.ai/v1/chat/completions \
-H "Authorization: Bearer $MISTRAL_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "mistral-small-latest",
"messages": [{"role": "user", "content": "Hello, Mistral!"}]
}'{
"object": "chat.completion",
"model": "mistral-small-latest",
"choices": [
{
"index": 0,
"finish_reason": "stop",
"message": {
"role": "assistant",
"content": "Hello! How can I assist you today?"
}
}
]
}Mistakes to avoid
The import path changed in SDK version 2; with mistralai 3.7.0 the old import fails with ImportError (our local check).
Browsers cannot keep long-lived keys safe.
The docs list only 429, 500, 502, 503 and 504 as transient.
Mistral Large 3 (mistral-large-2512) is deprecated, with Mistral Large 4 as its replacement.
The OCR service has to fetch the document itself.
Questions
Is there a free tier for the API?
No. The free plan does not include API access; creating keys and calling the API need billing activation and pay-as-you-go.
Who owns the text the models write for me?
You do. The commercial terms say the customer owns all output.
Can I keep my prompts out of model training?
Yes. An admin turns off the anonymous improvement data toggle in the Admin Panel's Privacy menu.
Is it hard to move from OpenAI?
Usually not. The chat request has the same structure, so you change the client import, the setup call and the model name.
Which languages have official packages?
Python and TypeScript.