
Mistral AI Startup Credits: $1,000 in API credits
Open-weight AI models with commercial API, fast, efficient, and multilingual LLMs from Europe.
Premium: $79/year for unlimited deals
Already have an account? Log in
Deal Highlights
Our Mistral AI listing: $1,000 in API credits, a Premium deal
Offer: $1,000 in API credits. Requirement in our listing: AI startup, especially EU-based. There is no promo code. This is a Premium deal on SaaSOffers, so the claim link sits behind the Premium upgrade.
We checked these plans and prices against Mistral AI's official pricing and help pages in November 2026.
Mistral AI is a Paris company that sells language models two ways. You can call them through a hosted API (the developer product is now called Studio, and the old name La Plateforme still turns up in older articles), or you can download the open-weight ones and run them yourself. The company also sells Le Chat and Vibe, the chat and coding apps for people rather than developers.
The $1,000 figure is our listing's term. Mistral's pricing page, its billing documentation and its help center do not mention a startup credit program or a $1,000 grant. We could not confirm the amount, the eligibility rules or an expiry date from Mistral's own pages. Treat the credit as unconfirmed until Mistral tells you the number in writing, and ask what it covers. Everything below about prices comes from Mistral's pages. Everything about the credit is "if it applies to your account."
What the credit would buy at Mistral's current API prices
Mistral's pricing page gives one API price in plain numbers: "Mistral Large costs $0.5 /M tokens in and $1.5 /M tokens out." That is $0.50 per million input tokens and $1.50 per million output tokens. The same page says batch processing is 50% cheaper and cached input tokens can cut input cost by up to 90%. OCR is priced per 1,000 pages, speech models per minute and tool APIs per call, and the page gives no dollar rates for those.
The model card for Mistral Small 4 lists "$0.15 and $0.6" per million tokens. The card does not label them. The order suggests input and output, but that is our reading, not Mistral's statement. Small 4 is Apache 2.0 licensed, has a 256k-token context window and is listed at 119B total parameters with 6.5B active.
Prices for Medium, Codestral, Devstral, Voxtral, embeddings and fine-tuning were not on the pages we could read. We left them out. Check the model cards in the docs before you budget for them.
Worked example 1: a support bot
Say your product sends 100 million input tokens and gets 20 million output tokens back each month.
- On Mistral Large at $0.50 and $1.50: 100 x $0.50 = $50, plus 20 x $1.50 = $30. Total $80 a month.
- Through the batch discount, if your jobs can wait: 50% of $80 = $40 a month.
- On Small 4, if the two numbers are input and output: 100 x $0.15 = $15, plus 20 x $0.60 = $12. Total $27 a month.
If our $1,000 applies, it covers 12.5 months of the Large figure ($1,000 / $80), or 37 months of the Small 4 figure ($1,000 / $27). Billing is not that clean, though. Mistral's billing page says credits are deducted including tax, while spending limits apply before tax, so a credit balance runs down slightly faster than the raw token math.
Worked example 2: a heavy pipeline
Now a document pipeline: 1 billion input tokens and 200 million output tokens a month.
- Mistral Large: 1,000 x $0.50 = $500, plus 200 x $1.50 = $300. Total $800 a month.
- With the batch discount: $400 a month.
If our credit applies, $1,000 lasts 1.25 months at the full rate, or 2.5 months in batch. For a workload like this the credit is a long test run, not a year of free usage.
The free tier and the rate-limit ladder
New Studio accounts start in free mode. Mistral's help center says free mode "has the lowest limits, intended for evaluation and prototyping." It does not publish numbers for the free limits. Rate limits are set per organization and applied per model, on three measures: requests per second, tokens per minute and tokens per month.
To go past free mode you enable pay-as-you-go, which needs prepaid credits. The help center then lists tiers by cumulative billed amount:
| Tier | How you reach it |
|---|---|
| Free mode | Default for new accounts, evaluation limits |
| Tier 1 | Enable pay-as-you-go |
| Tier 2 | Over 20 EUR or USD billed |
| Tier 3 | Over 100 EUR or USD billed |
| Tier 4 | Over 500 EUR or USD billed |
| Higher limits | Over 2,000 EUR or USD billed, then contact support |
Prepaid credits do not change your tier. Mistral says tiers rise only as you use tokens and reach the billing thresholds. A $1,000 credit that you burn through will push you up the ladder, but just holding the balance will not.
How prepaid credits behave
Mistral's billing page sets these rules for credits.
- You buy credits in advance with "Add credits," or redeem a gift code. Credits are required to turn on pay-as-you-go.
- Auto Recharge charges your payment method and adds a set amount when your balance falls below a threshold you choose.
- Your plan's included monthly usage is consumed first. After that, API usage needs pay-as-you-go and a positive balance.
- When the balance reaches zero, API requests beyond included usage fail with HTTP status 402.
- Admins and billing-role members get an email when 80% of the last top-up is used.
- A monthly spending limit, set at the organization level, can suspend API access until the next month or until an admin raises it.
- The balance can dip slightly below zero, and the next invoice includes the overage.
The page shows an expiration and status column in the credit history, but it does not state any expiry rule. That is the first question to ask about a promotional credit.
Refunds follow a separate rule. Website subscriptions are refundable only if you cancel within 14 days of the first purchase. For API credits and wallet balance, Mistral support reviews requests case by case, for example accidental top-ups or duplicate charges. Support asks for your organization ID and a receipt number.
Le Chat and Vibe plans are a different bill
The API is not the only thing on Mistral's pricing page. These are the plans for the apps:
| Plan | Price | Notes |
|---|---|---|
| Free | $0 | Limited messages, web searches and coding sessions, basic image generation, 100+ connectors |
| Pro | $14.99 a month, excluding taxes | Includes $15 a month in API credits, chat and email support |
| Education | $14.99 list, $5.99 for verified students | Student price valid up to 12 months, accredited higher-education students who have not used Vibe or Le Chat before |
| Team | $24.99 per user a month, excluding taxes | Up to 30GB storage per user, domain verification, data export |
| Enterprise | Contact sales | Custom models, agents and workflows, audit logs, SAML SSO, white labeling |
The page does not say whether those prices are billed monthly or yearly beyond the "a month" label, and it shows an unexplained "$50 /mo" figure next to Users on the Team card, so we did not use it. The subscription documentation says plans apply across Vibe, Studio and the API, with a monthly usage allowance shared between them. It does not give the allowance amounts.
The $14.99 Pro plan includes $15 of API credits each month, which covers light prototyping without any grant. Our $1,000 offer matters for people building products on the API.
Models that changed since older articles
If you read an older Mistral guide, parts of it are out of date. The model overview in Mistral's docs now lists these models as available, with licenses as stated there:
- Mistral Small 4 (Apache 2.0)
- Mistral Large 4 (Mistral Ecosystem Open Weight License)
- Ministral 3 in 14B, 8B and 3B sizes (Apache 2.0)
- Voxtral Small (Apache 2.0), plus Premier-tier Voxtral transcription and speech models
- Codestral and Codestral Embed (Premier), Mistral Embed (Premier)
- OCR 4.1 and OCR 3 (Premier)
- Shieldstral 1.0 (Apache 2.0) and Mistral Moderation 2 (Premier)
- Third-party models hosted on the platform, such as Z.ai GLM 5.3 and DeepSeek V4.1 Flash (MIT)
Mistral retires models on a schedule. The deprecation table shows Mistral Large 3 deprecated on October 31, 2026 and retiring on April 30, 2027, with Large 4 as the replacement. Mistral Small 3.2 retired on July 31, 2026. Mistral Medium 3.1 retired on August 31, 2026.
Do not hard-code a model name you found in a tutorial. Pin the version you tested and put the retirement dates in your calendar.
Open weights, licenses and self-hosting
Mistral publishes open weights for several of its models, so you can run them yourself. Small 4, the Ministral 3 family and Voxtral Small are Apache 2.0, which allows commercial use. Mistral Large 4 is open-weight too, but under the Mistral Ecosystem Open Weight License. Read that license before you build a business on it. We did not review its terms, so do not assume it matches Apache 2.0.
Self-hosting changes the cost model. You pay for GPUs instead of tokens. Small 4 is a 119B-parameter model with 6.5B active, so memory needs follow the total size. Whether self-hosting beats $0.15 per million input tokens depends on your volume and on how well you keep GPUs busy. We have not benchmarked it, so test before you commit.
Data handling, privacy and the EU angle
The opt-out article in Mistral's help center says that in the Admin panel, under Privacy, the "Anonymous improvement data" toggle controls whether API calls and related data are used to improve Mistral's services. The Vibe and API toggles are separate. The article does not state the default for Studio and API accounts, so check the toggle on day one instead of assuming.
Mistral is headquartered in France, and its Trust Center at trust.mistral.ai covers security, compliance and data protection. We did not find a page that states where API requests are processed, so we cannot confirm EU-only processing. If your enterprise customers ask for it, request it in writing along with a data processing agreement, and ask about zero data retention. We could not confirm from Mistral's own pages whether zero data retention is offered or on which plans.
The requirement in our listing says "especially EU-based." Mistral's own pages do not describe an EU-only startup condition. Treat that as part of our listing's terms.
Mistral against three other API providers
| Provider | Verified entry price point | Free option | Page on SaaSOffers |
|---|---|---|---|
| Mistral AI | Large $0.50 in / $1.50 out per 1M tokens; Small 4 $0.15 / $0.60 (order inferred) | Free mode with evaluation limits, no numbers published | This page |
| Groq | GPT-OSS 120B $0.15 in / $0.60 out per 1M tokens | Free plan: 30 requests per minute, 1,000 requests a day on GPT-OSS 120B | Groq |
| Anthropic | Claude Haiku 5.5 $0.10 in / $0.50 out (prompts up to 100K tokens; $0.50 / $2.50 above); Sonnet 5.5 $2 / $10 (per 1M tokens) | Free tier details not checked for this page | Anthropic |
| Cohere | Embed 5 Fast $0.08 per 1M text tokens | Free tier details not checked for this page | Cohere |
These providers do different jobs. Groq runs open-weight models, including GPT-OSS, on its own chips. Anthropic's models are closed and sit at the high end of the price range. Cohere is strongest on embeddings and reranking. Mistral is the one in this group that offers a general chat model, open weights you can download and a French jurisdiction in one account.
On price alone, Anthropic's Haiku 5.5 input rate ($0.10 for prompts up to 100K tokens) is lower than Mistral Large's ($0.50) and Small 4's ($0.15) in the figures above, so do not assume Mistral is the cheapest. Compare on your real prompts and check the output rates too.
Who it is for
- Teams that sell to European customers and want a European model vendor in the security questionnaire.
- Products that need an open-weight fallback, so a vendor change does not stop the app.
- Document-heavy products that need OCR, since Mistral lists OCR 4.1 and OCR 3 as models with per-1,000-page pricing.
Who should skip Mistral AI
- Teams that need a published startup grant with a written amount. Mistral's pages do not show one, so the $1,000 stays unconfirmed.
- Teams that want a firm EU-only processing guarantee before they have a contract. We could not confirm it.
- People who only want a chat assistant for themselves. The $14.99 Pro plan or the free Le Chat plan covers that without our offer.
Mistral AI setup tips
- Create a Studio account and stay in free mode while you test on real prompts.
- Turn off "Anonymous improvement data" under Privacy if your data is sensitive.
- Set a monthly spending limit at the organization level before you turn on pay-as-you-go.
- Set Auto Recharge to a small amount, so a bug does not drain a large balance.
- Handle HTTP 402 in your code. It is what you get when credits hit zero.
- Pin model versions and track the deprecation table.
- Send 200 real inputs to Mistral and to your current model, and compare quality and cost.
How to get the Mistral AI offer
- Upgrade to SaaSOffers Premium, then open this deal's claim link on SaaSOffers.
- Create or sign in to your Mistral account and follow the instructions on the claim page.
- Have your startup's details ready. Our listing says the offer is for an AI startup, especially an EU-based one.
- There is no promo code to enter.
- Before you rely on the credit, ask Mistral in writing for the amount, what it covers and when it expires. Billing starts from your prepaid balance and spending limits, so check them in the Admin panel.
If Mistral confirms a different amount than $1,000, go with Mistral's number.
Who Is This Deal For?
Early-Stage Startups
Seed and pre-seed companies looking to move fast without overspending on tools.
Growing SaaS Teams
Series A+ companies scaling their stack and optimizing software costs.
Solo Founders
Indie hackers and bootstrapped founders who need enterprise tools at startup prices.
Get $1,000 in API credits off Mistral AI
Premium deal. Upgrade once, unlock everything.
!Eligibility Requirements
AI startup, especially EU-based
Frequently Asked Questions
Everything you need to know about this startup deal.
Get the weekly deals digest
New verified startup deals every week. No spam, ever. Unsubscribe anytime.
Related Offers
Replicate
Used by 1,772 members
$500 in credits
Run open-source ML models in the cloud, deploy Llama, Stable Diffusion, and custom models via API without GPU management.
View offerConsensus
Used by 1,544 members
Free Plan
AI-powered research platform that finds, summarizes, and synthesizes scientific papers.
View offerLaxis
Used by 2,176 members
Free Plan
AI meeting assistant that records, transcribes, and generates actionable meeting notes.
View offerZoomInfo
Used by 2,014 members
Free Trial
B2B data platform for sales and marketing intelligence with contact databases.
View offerDeal Summary
Looking for more startup deals?
Browse all offers