Mistral AIAI and LLM Providers
European Inference, or the Same Models Run Locally
Mistral is a European company publishing both commercial and open-weight models, with EU-hosted inference. It is the provider named when a residency requirement is European rather than American, which is a different question from whether the data is encrypted or who holds the contract, and one that a US-hosted provider cannot answer however good its terms are.
What you can do together
- Keep inference in the EU, where that is the requirement rather than a preference.
- Run Mistral's open-weight models locally through the self-hosted runtimes, with no external call at all.
- Move from hosted to self-hosted within the same model family, without rebuilding workflows.
- Use Mistral for one agent and another provider for a different one, per sensitivity.
How it connects
The model behind an agent or a node is configuration. Mistral is selected per deployment or per agent.
The useful property is that Mistral exists twice over. Its commercial models run on Mistral's EU-hosted API, and its open-weight models can be run on the organization's own hardware through the self-hosted runtimes the platform already supports. So a deployment can start on the hosted API and move the same model family in-house later without changing what the workflows do, which is rarely true of a provider.
An engagement settles which of the two is in scope, because they answer different requirements. EU hosting satisfies a residency rule while still being an external call. Self-hosting removes the call entirely, and is what an air-gapped deployment needs. Choosing on price alone tends to pick the wrong one.
Mistral AIAI and LLM Providers
A scenario
- ScopingThe engagement establishes whether EU hosting satisfies the obligation or whether no external call is permitted at all.
- SetupThe EU-hosted API is configured as the provider in AI Intelligence Hub.
- In useAnswers are generated in the EU, with retrieval running under the asking user's identity.
- A tighter unitOne department's material may not leave the network, so an open-weight Mistral model is run locally for that agent alone.
- UnchangedBoth agents use the same workflows; only the provider setting differs.
What stays where
Mistral supplies the models
The hosted API, the model catalogue and the licensing stay with Mistral.
EU hosting is not the same as no external call
One satisfies residency, the other satisfies isolation, and the engagement names which is required.
Open weights can come in-house
The same family runs on the self-hosted runtimes, so the choice is reversible.
Content governance does not move
Retrieval runs under the asking user's identity whichever provider generates the answer.
Products and solutions
Next step
See it on your own Mistral AI instance.
We will show the connection made, the data moving and the output, then size it for your deployment.
Contact VIDIZMO
sales@vidizmo.ai
+1 571-969-2180
vidizmo.ai