Mistral AI deployment and training in Lithuania
European AI for organisations that care where their data goes. We deploy Mistral AI in the cloud or on your own servers and train your team to use it.
LONGRUN deploys Mistral AI and runs Mistral AI training for companies and public sector organisations in Lithuania. We help you choose the right deployment option, configure data protection settings and connect company documents and systems, and we train your staff to use Mistral in their daily work. Training and deployment are led by Audrius Grinys, founder of LONGRUN. He has worked with business management systems and company digitalisation since 2018 and deploys AI solutions for companies that handle sensitive information.
What is Mistral AI
Mistral AI is a French artificial intelligence company founded in Paris in 2023. It was founded by Arthur Mensch, Guillaume Lample and Timothée Lacroix, who previously worked in research teams at Google DeepMind and Meta. In September 2026 Mistral raised EUR 3 billion. The company is valued at more than EUR 21 billion. Its investors include ASML, Samsung, NVIDIA, BNP Paribas and the Grand Duchy of Luxembourg.
Main Mistral products:
- Mistral Vibe (called Le Chat until May 2026) is a chat assistant and work agent for employees, with connectors to company documents and systems.
- Mistral Studio and API is the developer platform used to connect Mistral models to company systems.
- Open models that can run on your own servers, for example Mistral Small 4 and Ministral 3 (Apache 2.0 licence).
Mistral is building its own data centres in France and plans to operate 200 MW of compute capacity in Europe by the end of 2027.
Why organisations choose Mistral
A European company under European law
Mistral is a European Union company, so the GDPR and the EU AI Act apply to it directly. Mistral has signed the European Commission's General-Purpose AI Code of Practice, which helps implement the AI Act requirements.
Data stored in the European Union
By default, Mistral stores customer data in the European Union. Enterprise customers can switch off features that would transfer data outside the EU. The developer API has a separate EU endpoint that processes requests in the EU region.
Deployment on your own servers
Mistral Vibe Enterprise can be deployed on your own servers, in a private cloud or in the Mistral cloud. Open Mistral models can run on your own infrastructure with no external provider at all. This suits organisations whose data must not leave their own infrastructure.
Your data is not used to train models
On the Enterprise plan, your data is not used to train models by default. On the Team plan, an administrator switches model training off. We check these settings together with you during training and deployment.
Certifications and contracts
Mistral holds ISO/IEC 27001 and ISO/IEC 27701 certifications and a SOC 2 Type II report. It signs a data processing agreement (DPA) with customers under the GDPR.
Named in Lithuanian cybersecurity recommendations
The recommendations of Lithuania's National Cyber Security Centre (November 2025) name Mistral AI among licensed large language models, alongside OpenAI and Anthropic. The same recommendations advise assessing whether a provider stores data in the EU, complies with the GDPR and the AI Act and holds ISO 27001 or SOC 2 certification. Mistral meets these criteria, and during deployment we prepare a provider assessment for your organisation.
Experience with the public sector and industry
Mistral has signed a framework agreement with the French Ministry of the Armed Forces, runs a dedicated public sector programme, "AI for Citizens", and works with SAP on sovereign AI solutions for European organisations. Mistral customers include Airbus, ASML, HSBC, BMW and CMA CGM.
What to know before you choose
Mistral is not the most capable model on the market. According to independent evaluations, the latest Mistral models trail the leading US models, such as ChatGPT, Claude and Gemini, on complex reasoning and coding tasks. In everyday work with documents, emails, summaries and internal search the gap is smaller, and the data security and legal clarity benefits often matter more.
Mistral rents part of its infrastructure from major cloud providers, including US companies whose data centres are located in Europe. If your data must not leave your own infrastructure, we recommend deploying on your own servers.
According to the vendor, the latest Mistral Large 4 supports all official EU languages. Before deployment we always test Lithuanian language quality on your real tasks, especially when smaller open models are used.
Who Mistral AI is for
- The public sector and municipalities that need an AI tool meeting EU data storage and provider assessment requirements.
- State and municipality-owned companies, energy, transport and other cybersecurity-regulated entities.
- Finance, insurance and healthcare companies that work with personal data and confidential information.
- Industrial and manufacturing companies that protect drawings, technologies and trade secrets.
- Lawyers, auditors and consultants who work with client documents and contracts.
- Organisations already using ChatGPT or Copilot that want a separate European tool for sensitive data.
Mistral AI deployment options
| Option | Best for | Where data is stored |
|---|---|---|
| Mistral Vibe Team | Teams that want to start quickly, without an IT project. | Mistral cloud, in the EU by default. |
| Mistral Vibe Enterprise | Organisations that need SSO, audit logs, administration and connectors to company systems. | Mistral cloud, a private cloud or your own servers. |
| Mistral API (EU endpoint) | AI integrations into existing systems: document processing, customer service, Odoo and other business systems. | EU region. |
| Open models on your own servers | Organisations whose data must not leave their infrastructure. | Your infrastructure only. |
What we do when deploying Mistral AI
- Assess your data and requirements, choose the deployment option and plan.
- Prepare a provider assessment against the NKSC criteria and GDPR requirements.
- Configure accounts, administration, data protection and model training settings.
- Connect company documents and systems and build agents for recurring tasks.
- Integrate Mistral via the API into your systems, for example Odoo, document management or customer service systems.
- Test Lithuanian language quality on your tasks and choose the right model.
- Prepare an internal AI use policy that sets out what is allowed, what is not and who is responsible for the result.
Mistral AI training programme
Topics are chosen to match your team's needs and participants' level. Each topic ends with a practical exercise using your own material. Training is delivered in Lithuanian or English, at your office or remotely.
Mistral Vibe basics
Chats, projects, working with files and web search. How to write a prompt that gets an accurate result and how to check the answers.
Documents, emails and summaries
Commercial proposals, replies to customers, summaries of long documents and contracts, document digitisation.
Agents and connectors
Building agents for recurring tasks, connectors to company documents and systems, shared team context.
Data security and settings
What data not to share in chats, model training settings, differences between plans, working under an internal AI use policy.
Administration (Team and Enterprise)
User management, access rights, audit logs, data protection settings for the whole organisation.
Mistral API and open models for the IT team
Mistral Studio, the EU endpoint, model selection, running open models on your own servers and integrations with company systems.
A practical task for your organisation
Each participant picks a task from their own work and, with the trainer's help, builds a solution they can keep using.
How it works
- Free consultation. We discuss your data, your requirements and the AI tools your team uses today.
- Choosing the solution. We choose the deployment option, plan and models and test Lithuanian language quality on your tasks.
- Deployment. We configure accounts and settings, connect documents and systems and prepare an AI use policy.
- Training. Staff learn in their own accounts on their own tasks; the IT team gets a separate technical session.
- After deployment. We review the results and, if agreed, maintain the solution and update agents and settings.
Mistral AI FAQ
Is Mistral AI data stored in the European Union?
By default, Mistral stores customer data in the European Union. Enterprise customers can switch off features that would transfer data outside the EU, and the API has a separate EU endpoint. If your data must not leave your infrastructure, Mistral can be deployed on your own servers.
Is Mistral AI suitable for the public sector?
Yes. Mistral is an EU company, complies with the GDPR, has signed the EU General-Purpose AI Code of Practice, holds ISO 27001 certification and a SOC 2 report, and is named among licensed language models in the NKSC recommendations. During deployment we prepare a provider assessment for your organisation.
Does Mistral AI speak Lithuanian?
According to the vendor, the latest Mistral Large 4 supports all official EU languages. Lithuanian quality varies across smaller models, so before deployment we always test it on your real tasks.
Is Mistral as good as ChatGPT or Claude?
On complex reasoning and coding tasks the leading US models are stronger. Mistral is the choice when data security, a European provider and the option to deploy on your own servers matter most. Many organisations use several tools: Mistral for sensitive data and other tools for general tasks.
How much does Mistral AI cost?
As of October 2026, Mistral Vibe Pro costs EUR 17.99 per month and the Team plan costs EUR 29.99 per user per month (minimum EUR 60 per month), prices for Lithuania including VAT. Enterprise pricing is available on request. The price of our deployment and training depends on the scope, and we send a proposal after a short call. Contact us at +370 637 37 637 or [email protected].
Can Mistral be deployed on our own servers?
Yes. Mistral Vibe Enterprise can be deployed on your own servers or in a private cloud, and open Mistral models can run on your own infrastructure. We help you estimate the hardware you need and which option pays off.
Does Mistral use our data to train models?
On the Enterprise plan, model training on your data is off by default. On the Team plan an administrator switches it off, and on the Free and Pro plans users can switch it off in their settings. During deployment we configure these settings for the whole organisation.
Is the training suitable for beginners?
Yes. Staff training requires no technical background. For the IT team we offer a separate technical session on the API, integrations and deploying open models.