Build AI workflows with Mistral in Knime

More choice in model providers means you can build AI workflows around what your project actually needs, whether that's a specific model's strengths, a certain price point, or where your data is processed. Workflows built in Knime aren't tied to a single provider, so as your requirements evolve, your work carries over.

KNIME Analytics Platform 5.12 widens that choice with dedicated support for Mistral. Three new nodes, the Mistral AI Authenticator, the Mistral AI LLM Selector, and the Mistral AI Embedding Model Selector, let you build AI workflows on Mistral's LLMs and embedding models.

The new nodes let you start working with Mistral in a few clicks. The Mistral AI Authenticator handles the connection to your Mistral account. Once authenticated, the LLM Selector and the Embedding Model Selector let you choose a Mistral language model or embedding model and use it in your workflow, the same way you would with any other provider in Knime.

If you've used the OpenAI or Gemini nodes in Knime, everything will feel familiar. The Mistral nodes share the same structure, so moving an existing workflow to Mistral mostly means exchanging a few nodes. Your prompts, your data processing, and the rest of your workflow stay as they are.

For many organizations in Europe, the choice of AI provider isn't purely a technical decision. Data protection officers ask where prompts are processed. Procurement asks which jurisdiction the provider falls under. In regulated industries such as banking, insurance, healthcare, and the public sector, these questions can stall an AI project for months.

Mistral is headquartered in Paris and operates under EU jurisdiction, which makes those conversations considerably shorter. If your compliance requirements point toward a European provider, you can now meet them in Knime without giving up anything in how you build.

To be clear about what this does and doesn't mean: using Mistral's API still sends your data to an external service. If your requirements rule that out entirely, Knime's support for locally hosted models remains the right answer. But if your rules allow a cloud API from a European provider, you're covered now.

You can ask questions across entire data tables, build AI assistants that answer from your own documents and data, or set up AI agents that carry out multi-step tasks in your workflows. If a task in Knime can use an LLM or an embedding model, it can now use one from Mistral.

Mistral's lineup covers a range of model sizes and price points, so you can match the model to the task, from bulk classification to complex reasoning.

KNIME Analytics Platform 5.12 is available now. The new nodes are part of the KNIME AI Extension, free to install, that brings LLMs and embedding models into Knime. Install or update the Knime AI Extension, get an API key from Mistral, and you're ready to go.