AI news · 2 min read

Mistral can host your workflow worker, with a preview limit

Mistral’s October 9 Managed Deployments preview runs workflow workers from GitHub. Paid access and concurrent deployment caps matter before a trial.

Announcement date:

Affiliate disclosure: Some links may earn CreatorStack a commission at no extra cost to you.

At a glance

Mistral’s October 9 Managed Deployments public preview builds and runs workflow workers from GitHub. It requires a paid plan: Pay-as-you-go allows three concurrent deployments and Enterprise ten. Check the workspace’s costs and limits, then validate one small workflow before relying on hosted execution.

An illustrated code review with branching changes and a checklist
CreatorStack illustration. Editorial illustration of this story’s topic, rather than a product screenshot or tested result.

An automation that works on your laptop still needs somewhere to run. Mistral’s October 9, 2026 Managed Deployments release lets its cloud operate a workflow worker from your GitHub code. It is a public preview requiring paid access, rather than a finished free hosting offer. Studio release notes.

You supply the workflow; Mistral runs the worker

Mistral builds an image from the repository’s Dockerfile and runs the container. The worker uses a service account instead of a workspace API key. Deployment lifecycle controls and build/worker logs are available through Studio, with lifecycle operations also exposed through the API. Worker mechanism, release capabilities.

For someone building a repeatable content-processing job, this moves server operation out of the immediate task. It does not remove the work of deciding what the automation should accept, produce and retry. A useful pilot is one small input with a clearly inspectable output, followed by a deliberately invalid input to see whether the failure is understandable. This is a proposed evaluation, not a deployment we ran.

Count running deployments before adding another

The organization cap is three concurrent managed deployments on Pay-as-you-go, ten on Enterprise and zero on Free. Creating or starting beyond the cap fails with a 403 error. These limits count running deployments, not completed jobs. Mistral chooses worker instance size, and per-deployment CPU and memory metrics are not exposed yet. Current preview quotas.

The reviewed release and quota pages do not quote a hosting tariff. Confirm your workspace’s billable services and current limits before a pilot; do not read the quota as an included execution allowance.

Next step: verify the local workflow’s output and error behaviour before moving it to a hosted worker. Our guide to agent state, recovery and evaluation can help identify the failing step before infrastructure makes it harder to see.

Common questions

Questions people ask.

Does the three-deployment limit mean three workflow runs?
No. The published Pay-as-you-go cap counts managed deployments running concurrently. It is not a promise about the number of workflow executions or their throughput.
Can a Free organization use Managed Deployments?
The quota table gives Free a cap of zero. This public preview requires Pay-as-you-go or Enterprise access.

Sources & freshness

This article was refreshed on 10 Oct 2026 using the sources below. A source check verifies published information, not output quality or checkout availability in every region.

What changed: New coverage of managed workflow-worker hosting, its Dockerfile mechanism, paid preview access and organization concurrency caps.

Found a correction? Let us know.

Need help applying this? Explore a workflow audit or get future guides by email.