Products

AI Adoption & Privacy

AI Workspace
Pricing

Network Security Pricing

Network DDoS Protection
Partners
About Myra
Login
WAAP (Myra App) AI Workspace EU CAPTCHA Partner Portal
Deutsch English Français
Emergency Demo
Products
Application Security DDoS Protection Web Application Firewall CDN Bot Management CAPTCHA Network Security Network DDoS Protection (cloud) Network DDoS Protection (on-prem) Network DDoS Protection (hybrid)

AI Adoption & Privacy

AI Workspace
Pricing

Application Security Pricing

DDoS Protection, WAF, CDN, Bot Management EU CAPTCHA

Network Security Pricing

Network DDoS Protection
Partners
About Myra

Company

About Us Careers News & Insights Events 2026 Press Contact

Why Myra?

References & Case Studies Certifications Migration Integrations GDPR Compliance Digital Sovereignty

Myra Compared

Myra vs. Competitors Cloudflare Alternative Akamai Alternative F5 Alternative
Login WAAP Login AI Workspace Login EU CAPTCHA Partner Portal Deutsch Français
Emergency Demo

Updated 11 October 2026

Mistral Le Chat or Myra AI Workspace? A sourced comparison

As of 11 October 2026

Mistral Le Chat and Myra AI Workspace both come from European companies, so origin does not separate them. The approach does. Mistral builds its own language models, and its assistant works with exactly those models. Myra AI Workspace is a workspace where models from several vendors sit side by side, and each project decides which of them may see your data.

First, the name: Le Chat is now called Vibe. Mistral moved existing accounts and their conversations across, and the address chat.mistral.ai stayed the same, so this page says Vibe from here on. All statements about Mistral are sourced from Mistral's own pages.

What Mistral does well

Mistral does its own research and publishes many models with open weights, among them Mistral Small 4 and Mistral Large 3 under the Apache 2.0 licence. If you want to run the models in your own data centre, you can. Mistral also offers the assistant itself for self-hosting on its Enterprise plan, in your own cloud or on premises. For organisations that want nothing to leave the building, that is a strong argument. Myra AI Workspace runs as a service on Myra's infrastructure by default, with on-premises deployment on request.

According to its help centre, Mistral stores data in the EU by default. Data is hosted in the United States when a customer explicitly chooses the US API endpoint, and some features may temporarily move it outside the EU through sub-processors. Those sub-processors are listed publicly in Mistral's Trust Center, with location and product for each entry. Mistral operates its own infrastructure in France (Mistral Compute) and, for Vibe, also lists Microsoft Azure in Sweden and Norway, Google Cloud and CoreWeave. As certifications, it names ISO/IEC 27001, ISO/IEC 27701 and SOC 2 Type II.

On the Enterprise plan, customers are excluded from model training by default, and administrators can filter an audit log directly in the admin interface. If you want an assistant straight from the model maker, with the option to run it yourself later, Mistral is the obvious candidate.

Vibe and Myra AI Workspace on twelve points

Criterion Mistral Vibe (according to Mistral) Myra AI Workspace
Company Mistral AI SAS, Paris; Munich hub since September 2026 Myra Security GmbH, Munich
Models in the assistant Mistral models in every mode; the Vibe documentation does not describe models from other vendors Myra models on Myra's own EU infrastructure, currently including Mistral Small; external models from other vendors once an admin allows them¹
Self-hosting Enterprise: self-hosted, private cloud or on premises; many open-weight models Service on Myra infrastructure; on-premises on request; optional storage in Switzerland, with inference on Myra's EU infrastructure
Where data goes EU by default; US when the US API endpoint is used; some features use sub-processors outside the EU, web search for example EU routing can be enforced per gateway or for the whole organisation; not enforced out of the box
Infrastructure Mistral Compute (France), Microsoft Azure (Sweden, Norway), Google Cloud, CoreWeave; further providers for Vibe Code Web (e.g. Scaleway, OVHcloud) Myra infrastructure with a BSI C5 Type 2 attestation
Training on customer data Free, Pro, Team: on by default, opt-out at any time; Enterprise: off by default. Thumbs ratings with a comment on any plan Myra does not use customer data to train AI models; for external models, the provider's terms apply
Locking models per project Projects group chats with their own instructions and files; a per-project model lock is not described. Admins switch connectors and features off per workspace Access tier “Local only”: Myra models alone; cloud models and web search blocked, external tools too¹
Personal data Masking before the model is not described in the Vibe documentation Reversible masking before external models, with “Preview”; available in every plan
Audit log Enterprise: sign-ins and settings changes, user management, created and deleted conversations; filterable view for admins, no export according to the documentation Admin changes and security events are logged, tamper-evident via a hash chain on request; customer admins have no view of their own, queries run via API through Myra
Certifications ISO/IEC 27001:2022, ISO/IEC 27701:2019, SOC 2 Type II; Mistral does not list a BSI C5 attestation Myra Security: ISO 27001 on the basis of IT-Grundschutz; Myra infrastructure with a BSI C5 Type 2 attestation
Sub-processors Public list in the Trust Center, changes by email subscription No public list at present; Myra announces new sub-processors 30 days ahead, with a right to object
Chat without a trace “Temporary Chat”: not saved to history, not used for training; Mistral records only that it took place “Ghost mode”: nothing is stored, and the conversation stays out of the logs

¹ On paid self-service plans, the Myra-hosted models require the “EU-Gov” add-on.

We left prices out: Mistral charges per user with usage limits, Myra per user plus model usage by consumption, so the numbers do not compare directly.

Training defaults depend on the plan

For data protection officers, this matters most. According to its help centre, Mistral uses Vibe inputs and outputs to train its models by default unless someone opts out. On the Team plan an administrator can switch training off for the whole organisation, and Enterprise customers are excluded from the start. Mistral says so openly and puts the switch in the settings.

Two details still belong in your review. Anyone who rates an answer with thumbs up or down and adds a comment allows Mistral, on any plan, to use the rating together with the input and output to improve its models. And the opt-out for Vibe and the one for the API are separate switches. Data that Vibe pulls through connectors such as Gmail or SharePoint is never used for training, according to the documentation; conversations that use such data follow the plan's training rules.

In Myra, this does not depend on the plan. Myra does not use customer data to train AI models. If you allow external models, check their terms.

One model maker or a choice of several

Vibe works with Mistral models, which is consistent: Mistral sells its own technology. On using other vendors' models, the Vibe documentation says nothing.

Myra models run on Myra's own EU infrastructure, and Mistral Small is currently one of them. Admins can also allow external frontier models from other vendors, Anthropic for example. The project decides which model sees which data. Set its access tier to “Local only” and the model picker offers Myra models alone; cloud models, web search and external tools are blocked, and the server rejects cloud requests. Only the project owner can change the tier, with four-eyes approval if you want it. Legal can then review contracts in a local-only project while sales has an external model work on proposal texts in the project next door.

Before anything is sent, a coloured indicator next to the model picker shows what will happen to the text. Green, “Local model” or “Local + masking”, means the request stays with Myra models. Yellow, “Privacy on”, means an external model answers, and Myra first replaces personal data such as names or IBANs with placeholders that turn back into the original values in the answer. Red, “Privacy off”, means the external model receives the text unchanged. Detection fully supports German and English; for other European languages, use “Preview” to see what gets replaced. Under the GDPR, masking remains pseudonymisation.

With Vibe the masking question looks different, because Mistral's own models answer. There, control comes from the storage location, the training opt-out and which connectors an admin allows.

Who can read the log

On the Enterprise plan, Mistral records sign-ins, changes to the organisation and its settings, user management, and conversations being created or deleted. Admins filter the log in the interface; according to the documentation it cannot be exported. Mistral's audit-log documentation does not say how long entries are kept or whether changes to the log would be detectable.

Myra has it the other way round. Admin changes and security events are logged, tamper-evident via a hash chain on request, and kept for 365 days by default. Your own admins, however, have no view of the log; Myra retrieves extracts via API. If your internal audit team wants to browse the log itself, Mistral Enterprise is more convenient today.

Data centres and attestations

Both companies operate their own infrastructure. For Vibe, Mistral's Trust Center also names Microsoft and Google as well as CoreWeave as infrastructure providers. Myra Security is listed by Germany's Federal Office for Information Security (BSI) as a qualified DDoS mitigation service provider, and AI Workspace runs on Myra infrastructure that holds a BSI C5 Type 2 attestation. C5 is the BSI's catalogue of security criteria for cloud services; the attestation covers the infrastructure, and the application itself is outside its scope. Mistral does not list a C5 attestation, but it does hold ISO/IEC 27701 for privacy management.

Which situation points to which product

Your situation What fits better
The assistant has to run in your own data centre Both: Mistral Enterprise for self-hosting, Myra on-premises on request
You want models straight from the maker's own research Mistral
Internal audit wants to filter the audit log in the interface Mistral (Enterprise)
You want to read the sub-processor list without asking Mistral
No training on your data, without anyone having to flip a switch Myra, or Mistral's Enterprise plan
Some projects may only use self-hosted models, others need an external frontier model Myra (“Local only” per project)
Models from several vendors in one interface Myra
You require a BSI C5 attestation for the infrastructure operator Myra

If your organisation is bound by professional secrecy, ask both vendors what they offer contractually. For Germany, Myra offers a confidentiality undertaking under § 203 StGB and § 43e BRAO; we found nothing comparable on Mistral's pages.

Further reading: Myra AI Workspace, the comparison Langdock alternative and the German version of this page, Mistral Le Chat oder Myra AI Workspace?

All statements about Mistral come from Mistral's own pages and documents (see Sources), as of 11 October 2026. Mistral, Le Chat and Vibe are trademarks of their respective owners. Myra Security has no affiliation with Mistral AI.

Frequently asked questions

Is Mistral Le Chat now called Vibe?

Yes. According to Mistral's help centre, Le Chat has become Vibe, an agent for professional productivity and coding. Accounts, plans and conversations moved across, and chat.mistral.ai is still the address.

Is Mistral a European company?

Yes. Mistral AI is a French SAS headquartered in Paris, with a German subsidiary.

Does Mistral use my prompts for training?

In Vibe, yes by default, until you opt out in the settings. On the Team plan an administrator can switch training off for the whole organisation, and Enterprise customers are opted out from the start, says Mistral. Thumbs-up or thumbs-down ratings with a comment may be used for training on any plan.

Can I use Mistral models in Myra AI Workspace?

Yes. The Myra models currently include Mistral Small, an open-weight model that Myra runs on its own EU infrastructure. On a paid self-service plan, the Myra-hosted models require the “EU-Gov” add-on.

How do I test both without much effort?

You can try Vibe on its free plan straight away; switch off training in the settings first. With Myra AI Workspace you start a pilot in your own tenant, with SSO and a data processing agreement. Give both the same real tasks from your daily work.

Sources