Plans and Deployment

Update November 2, 2026 — Le Chat is now Vibe. As of May 28, 2026, Le Chat has been rebranded as Vibe (still at chat.mistral.ai): a single agent for work and code that carries over your conversations, settings, and subscriptions. The legacy Le Chat experience remains available in Chat mode. In Work mode, agents are now skills (invoked with @), memories serve as the knowledge base, and deep research is the Deep Research skill. The plans are now called Mistral Free, Pro, Team, and Enterprise. This lesson uses the name Vibe; note that some menus described here may have changed names or locations.

Match the plan to your organization’s maturity

Mistral AI offers three Vibe plan tiers. Your choice determines available features, support levels, and security guarantees—and, crucially, what you can promise your compliance leadership. A poorly calibrated plan costs you in two ways: too light, and it blocks production the day your DPO asks for a retention policy; too heavy, and you pay for enterprise-grade governance that a ten-person team doesn’t need.

Mistral Pro targets independent professionals and small teams that want top-tier models without enterprise constraints. You get access to all models—Mistral Large 3, Mistral Medium 3.5, Codestral, Pixtral—along with custom agents, Canvas, file uploads, and web search. However, there is no SSO, no admin console, and retention remains standard: data is used to improve the service, with an opt-out option. If a small business deploys Pro for fifteen people, it has no way to track who uses what or revoke access except manually.

Mistral Team serves teams of 5 to 200 people that need collaboration and basic administrative control. It includes all Pro features, plus a shared workspace for agents and documents, basic member management (add, remove, assign roles), centralized billing, configurable retention, and priority email support. This is the plan for a mid-sized company’s marketing department that wants to share agents without launching a full IT project.

Mistral Enterprise meets the demands of large organizations and regulated industries. It builds on Team by adding SSO/SAML 2.0 and SCIM provisioning, zero-data retention by default, a full admin console with audit logs, granular role and permission management, access to Enterprise Search and internal connectors, a 99.9% availability SLA, dedicated support with a Customer Success Manager, and the option for on-premise or dedicated VPC deployment. Each of these items answers a specific question you will face during production rollout.

Three deployment models

The standard deployment is the Mistral cloud (SaaS): access via browser or mobile apps, with infrastructure managed by Mistral AI in European data centers, primarily in France and the Netherlands. Setup takes a few hours, requires no infrastructure investment, and includes automatic model updates and auto-scaling. This fits most organizations, including those subject to the GDPR—a point many committees underestimate, assuming regulated sectors always require dedicated infrastructure.

A dedicated VPC provisions an instance in an isolated cloud environment where your data never traverses shared infrastructure. You gain complete network isolation, dedicated IP addresses, custom network configuration (VPN or peering), and enhanced compliance. This model targets banks, insurers, healthcare providers, defense contractors, and government agencies with specific network requirements, such as mandatory IP range filtering.

On-premise deploys models directly into your infrastructure on your own GPU servers: no data leaves your network. You get total control, no external network dependencies, support for air-gapped environments, and maximum compliance. The trade-off is significant: substantial hardware investment (H100 GPUs or equivalent), an internal team for maintenance, manual model updates, and a four-to-twelve-week deployment timeline. Reserve this for contexts that truly demand it—defense, intelligence, or organizations requiring absolute data sovereignty.

Execute the deployment

Evaluation kicks off the project and takes one to two weeks: define priority use cases, audit existing infrastructure (SSO, network, security), estimate usage volume, and select the plan and deployment model.

Configuration takes another one to two weeks to set up SSO and user provisioning, define retention policies, create initial business agents, and configure the admin console.

Next comes the pilot, running two to four weeks with 50 to 200 users. You collect feedback, make adjustments, measure early adoption metrics, and train administrators and internal champions. Neglecting these champions is the most common mistake; they will field questions from colleagues for the next six months.

General rollout follows, expanding access progressively across the organization alongside internal communications and training, then monitoring adoption and optimizing continuously. Plan for six to ten weeks end-to-end.

Support included in the Enterprise plan

The Enterprise plan provides a dedicated Customer Success Manager as your single point of contact at Mistral, plus technical support with a guaranteed response time of under four hours for critical incidents. You also get quarterly reviews covering usage analysis, optimization recommendations, and product roadmap updates, along with training sessions for administrators and power users. Use these reviews to revisit your retention policy and the list of enabled models—two parameters that age quickly.

Key takeaways

• Three plans (Pro, Team, Enterprise) to match your organization’s size and requirements
• Three deployment models (SaaS, dedicated VPC, on-premise) based on your sovereignty needs
• SaaS deployment fits most organizations; reserve dedicated VPC and on-premise for highly regulated sectors
• The full deployment process, including the pilot, takes 6 to 10 weeks