Mistral AI components for chat completions, vision, embeddings, document OCR, speech-to-text, text-to-speech and content moderation.
Changelog
3 versions
v1.1.1
Latest
- AI Agent: the tool calls of one turn run at the same time (calls of the same Tool Start one after the other). A Tool Start that is the only kind of tool on the port is called through the port; next to other tools it is called through its webhook URL.
- AI Agent: a field of a tool mapped to the 'Prompt' variable of the 'tools' port gets the prompt the agent is answering. A Tool Output that comes after its tool call has timed out is removed from the flow state with the next message of the agent.
v1.1.0
- New AI Agent component: model and instructions on the input port (bindable to flow variables); tools on the 'tools' port (the same port as the OpenAI and Claude agents) - any action component (fields mapped to 'Model Defined Parameter' are filled by the model), a chain of components run as one tool, a 'Tool Start' with its own description and parameters whose chain may branch and ends with 'Tool Output', or an MCP server. Thread memory with an optional data store, files as context (images to vision models, PDFs through Mistral OCR) and streaming to the chat utility. The model list offers models that support function calling.
- Send Prompt: the output now has the same shape as the other AI connectors - Answer, Prompt, ID, Model, Created Timestamp, Finish Reason and Usage as top-level fields. The previous 'choices.message.content' and 'choices.finish_reason' outputs are replaced by 'answer' and 'finishReason'.
v1.0.0
- Initial release.
- Components: Send Prompt, Describe Images, Generate Embeddings, Extract Document Text, Create Transcription, Create Speech, Moderate, List Models, List Voices and Make API Call.
- Model dropdowns are filtered by model capability so each component only offers models the endpoint accepts.
Actions
11
AI Agent
Build an AI agent on a Mistral model that answers with context and acts through tools. Connect a component to the 'tools' output port to let the agent call it as a tool, and mark the fields the model should fill with the 'Model Defined Parameter' variable; components connected behind it run as further steps of the same tool. For a tool with its own description and logic (branches, loops), connect a 'Tool Start' to the 'tools' port and end every branch with 'Tool Output'. An MCP server on the 'tools' port gives the agent all of its tools.
Create Transcription
Generates a transcription of audio content to text using a Voxtral speech-to-text model.
Extract Document Text
Extracts text (as markdown) from a PDF document or an image using the Mistral Document OCR API.
Describe Images
Answers a question about the content of one or more images using a vision-capable Mistral model.
List Models
Lists the models available in your Mistral AI account, including their identifiers and metadata.
List Voices
Lists the text-to-speech voices available in your Mistral AI account, including presets and custom voices.
Moderate
Classifies text into content moderation categories (sexual, hate, violence, self-harm, PII and more).