Writing an AI workflow is one job. Keeping the software that runs it online is another. Mistral’s Managed Deployments preview takes on that second job: its cloud builds and runs your workflow code from a GitHub repository.

Based on 4 reviewed sources · See sources

Your code, Mistral’s servers

The 9 October release note describes a public preview for paid plans. A worker—the program that carries out a workflow’s steps—runs on Mistral Cloud rather than infrastructure the customer operates. Teams can manage deployments and read build and worker logs in Studio.

This still requires a software project. The setup guide supports public or private repositories on github.com; GitLab and GitHub Enterprise support are planned. Mistral builds a Docker container, a package containing the code and its dependencies, and runs it. A service account handles worker authentication without a worker API key.

A preview with a small deployment allowance

The quota documentation limits how many managed deployments an organisation can run at once. These are deployment slots, not a limit of three or ten workflow executions. Mistral chooses the worker’s instance size, and per-deployment CPU and memory metrics are not yet exposed.

The detail

Concurrent managed deployments

Public-preview caps per organisation

Concurrent managed deployments. Public-preview caps per organisation
PlanDeployments running at once
Free0 — paid plan required
Pay-as-you-go3
Enterprise10
Caps can be raised through Mistral support. These are concurrency limits, not a price quote; preview limits may change.Source: Mistral quota documentation · 2026-10-10

The worker runs in the Netherlands

Mistral’s architecture documentation places the managed worker in nl-north-1, in the Netherlands. That is a concrete European hosting detail. It does not by itself establish where every connected service processes data or make a workflow compliant.

The practical change is less infrastructure to operate, while teams retain responsibility for the workflow code and its connections. The documentation establishes access and limits; it supplies no independent evidence of faster execution or better business results.

Sources

Written with AI assistance from the sources above. Analysis and practical implications are our interpretation. Our editorial approach.