Mistral AI Pricing: Free, Pro, Team, API and Enterprise Plans

by David Porter

Friday, 28 August 2026 at 22:04

thumbnail_mistral-ai-pricing-free-pro-te

Mistral AI pricing has two main layers:

a Vibe subscription for the hosted Work, Chat and Code interfaces;
pay-as-you-go API billing for developers using models and tools through Mistral AI Studio.

They share an account, but they are not interchangeable. Paying for Vibe Pro does not create unlimited API access, while paying for API tokens does not automatically unlock every Vibe feature.

This guide uses Mistral’s public US-dollar prices on 21 August 2026. Taxes, currencies, regional availability and enterprise contracts can change the final amount.

Mistral Vibe plans at a glance

Plan	Public price	Main use
Free	$0	Testing Vibe, Studio and limited coding
Pro	$14.99/month	Regular individual work and all-day coding
Team	$24.99/user/month	Shared managed workspace for organizations
Enterprise	Contact sales	Private deployment, customization and advanced governance
Education	$5.99/month	Discounted Pro access for eligible students

Pro and Team can be billed monthly or annually. Mistral says annual billing includes a 20% discount compared with the monthly rate.

All public prices exclude taxes and paid plans remain subject to fair-use limits. Check the official pricing page immediately before purchase.

Mistral Vibe Free

Free is designed for evaluation and light personal use.

Mistral currently lists:

Vibe access on web and mobile;
limited messages and web searches;
limited coding sessions;
image generation;
access to test models in Mistral AI Studio;
more than 100 available connectors;
and $10 per month in API credits.

The Free plan can answer questions, work with files, search the web and give users a practical view of Vibe’s three interfaces. Read our complete Mistral Vibe guide before testing it.

Limits can affect sustained research, repeated image generation and long coding sessions. Free also has a different training default from the business plans: Mistral says Free inputs and outputs may be used for model training unless the user opts out.

Free is the sensible starting point for public or low-sensitivity tasks. It should not become an unofficial company workspace simply because no invoice is required.

Mistral Vibe Pro

Pro costs $14.99 per month, excluding tax.

It raises access across messages, web searches, longer tasks, image generation and coding. Mistral describes it as full Vibe access for all-day coding and long-running work.

Current Pro benefits include:

more messages and searches;
support for more complex agent tasks;
coding through the web, CLI and IDE;
more image generations;
up to 15 GB for libraries and uploaded material;
chat and email support;
and $30 per month in API credits.

Pro is aimed at one person rather than a managed company workspace. It fits a developer, analyst, researcher or independent professional who regularly reaches Free limits.

The $30 API allowance is a credit toward metered API use. It does not turn Pro into a flat-rate developer plan. Once credits are exhausted, enabled pay-as-you-go use is charged at the relevant API rates.

Free, Pro and Education users should also inspect their privacy setting. Mistral currently uses their Vibe inputs and outputs for training by default unless they opt out. Our guide to Mistral privacy and security explains the distinction between training, retention and operational processing.

Mistral Vibe Team

Team costs $24.99 per user per month, excluding tax.

It provides a collaborative workspace with organizational controls rather than a collection of unrelated Pro accounts. Mistral’s current plan page highlights:

up to 30 GB of storage per user;
domain-name verification;
data export;
shared workspace administration;
higher allowances for some Vibe features;
and company-level user controls.

Mistral’s help documentation describes administrator controls including user management and the ability to disable Flash Answers for the organization. The public package places SAML SSO in Enterprise. Confirm the exact boundary during procurement, especially for audit logs, identity lifecycle and support.

A material privacy difference appears at this tier: Mistral says Team inputs and outputs are not used to train its models. Users cannot turn training on for the workspace.

Choose Team when several people need governed access, shared context and central administration. An employer should prefer a managed workspace over reimbursing personal Pro accounts for work involving company information.

Mistral Enterprise

Enterprise uses negotiated pricing.

The public package is built around private deployments and customization. Mistral lists:

custom models;
custom agents and workflows;
audit logs;
SAML SSO;
white labelling;
private cloud or on-premises options;
dedicated support and custom service levels.

Enterprise is relevant when the deployment must fit an existing identity system, data architecture, security program or regulated environment. Price depends on more than seats. Model customization, infrastructure, reserved capacity, implementation, support and private deployment can all affect the contract.

Mistral says Enterprise data is excluded from model training. Default EU hosting and available private deployments can support European procurement requirements, but the customer still needs to review subprocessors, transfers, retention and the Data Processing Addendum.

For strategic context around the company, models and European positioning, use the Mistral hub.

Mistral Education

Mistral currently advertises an Education plan at $5.99 per month for eligible verified students. It is a discounted version of Pro rather than a separate feature family.

Eligibility conditions can include institutional verification, a limited discount period and account-history requirements. Confirm the current offer during checkout instead of treating $5.99 as a permanent universal student price.

Annual billing discount

Mistral’s help center says Pro and Team annual subscriptions receive a 20% discount compared with monthly billing.

At the displayed monthly rates, that implies an approximate pre-tax equivalent of:

Pro: about $11.99 per month when paid annually;
Team: about $19.99 per user per month when paid annually.

These are calculations from the stated discount, not separately quoted checkout prices. Currency conversion, tax and localized pricing can alter the final total.

Annual billing makes sense after the product has passed a real trial. Test the same recurring tasks on Free or one month of Pro first. Measure correction time, failed agent runs and whether Vibe replaces another paid tool.

Mistral API pricing is separate

Mistral’s API is billed by usage through Mistral AI Studio. The developer chooses a model and pays for inputs, outputs and any additional tools.

Our Mistral API guide covers authentication, model IDs and code. The table below shows selected standard rates published on 21 August 2026.

Model	Input per million tokens	Output per million tokens
Mistral Small 4	$0.15	$0.60
Mistral Large 3	$0.50	$1.50
Mistral Medium 3.5	$1.50	$7.50
Codestral	$0.30	$0.90

The appropriate model depends on the task:

Small targets cost-sensitive general workloads.
Large is the open-weight flagship general model.
Medium is optimized for long-running agentic and coding work.
Codestral is specialized for frequent code completion and generation.

See the Mistral models overview before choosing from an alias such as mistral-medium-latest. A “latest” alias can point to a newer implementation after this article is published.

Mistral also prices other services differently:

OCR and Document AI are billed per 1,000 pages;
speech services can be billed per audio minute or generated characters;
web search and code execution add tool-call charges;
image generation is billed per generated image;
Libraries can add OCR, indexing and call costs.

The live API pricing page should remain the source of truth for production estimates.

Batch, caching and regional inference

Headline token prices do not describe every API bill.

Mistral currently offers:

Batch processing and cached inputs: a 50% batch reduction and up to a 90% reduction on repeated eligible input tokens;
regional inference: supported EU or US processing options with a 10% surcharge;
Priority Tier: prioritized inference at 1.75 times the standard model price for entitled customers.

Caching can materially reduce the cost of repeatedly sending a long shared system prompt or document. Batch helps when immediate responses are unnecessary. Regional and priority pricing can be worthwhile where data location or latency is a contractual requirement.

Do the included API credits make Pro cheaper?

Free currently includes $10 per month in API credits, while Pro includes $30.

For a user who genuinely needs both Vibe and the API, those credits reduce the effective combined cost. They have little value if the user never builds or tests an API workflow.

Before subscribing, confirm:

whether credits expire each month;
which services they cover;
whether pay-as-you-go is enabled after they run out;
which spending limits and alerts are configured;
and whether several people are sharing the same organization budget.

Never expose an API key in browser code, public repositories or prompts. A surprise bill often starts with a leaked key rather than an expensive model.

Hosted API versus running Mistral locally

API cost should also be compared with self-hosting.

Open-weight Mistral models can run on customer-controlled hardware or infrastructure, but local inference still has costs:

GPUs, RAM and storage;
electricity and cooling;
serving software;
engineering and monitoring;
security, backups and patching;
idle capacity;
and model upgrades.

The local Mistral deployment guide explains which model sizes are realistic on consumer hardware.

A low API rate will often beat a lightly used private server. Self-hosting becomes more attractive when data control, predictable high utilization, latency or offline operation creates enough value to justify the operational burden.

Which Mistral plan should you choose?

Choose Free to test Vibe with public or non-sensitive tasks.

Choose Pro when one person regularly needs longer agent runs, higher limits or sustained coding.

Choose Team when an organization needs central administration, shared workspaces and a no-training commitment.

Choose Enterprise when private deployment, custom models, auditability, SAML or negotiated support is required.

Use API pay-as-you-go when software needs programmatic model access. A Vibe subscription remains optional.

Run open-weight models locally when deployment control outweighs the cost and operational work.

For a direct product comparison, see Mistral Vibe vs ChatGPT. Our wider ChatGPT alternatives guide compares Vibe with other assistants.

Frequently asked questions

Is Mistral Vibe free?

Yes. Vibe Free provides limited access to messages, web search, images, coding and Mistral AI Studio.

How much does Mistral Vibe Pro cost?

The public price is $14.99 per month before tax. Annual billing receives a 20% discount.

How much does Mistral Team cost?

Team costs $24.99 per user per month before tax. Mistral also offers annual billing at a 20% discount.

Does Mistral offer an Enterprise price?

Enterprise pricing is negotiated. Cost depends on deployment, customization, infrastructure, support and contractual requirements.

Is Mistral API access included with Pro?

Pro currently includes $30 per month in API credits. API use remains metered and separately billed after available credits.

Does Free include API credits?

Mistral’s current pricing page lists $10 per month in API credits with Free.

Which Mistral model is cheapest through the API?

Among the selected general models above, Mistral Small 4 has the lowest displayed rate at $0.15 per million input tokens and $0.60 per million output tokens. Smaller Ministral models may be cheaper for suitable edge-oriented tasks.

Is the Mistral API cheaper than self-hosting?

Often, for low or irregular usage. Self-hosting can become economical at sustained utilization or where privacy, latency and offline requirements justify the infrastructure.

Are Mistral prices the same in every country?

No guarantee. Currency, tax, regional availability and checkout pricing can vary. Verify the final amount for the billing country.

byDavid Porter

Massive Chinese Bot Farm Caught Sabotaging American AI Growth on X

OpenAI Leadership Crisis Deepens as Data Center Chief Suddenly Resigns

Write a comment