Mistral AI
Put Mistral AI to work on your own records, inside your rules, with a person approving what it changes.
36 ways to read, 9 ways to write, each write switched on by you.
Quality checks
Built and tested with example responses.
- Ready to connect: doneEvery listed action is built in.
- Built and tested with example responses: doneEach action is built from the provider’s published API and tested against example responses, errors, rate limits and paging.
Connected, approved, proven.
The things teams do most with Mistral AI, an example flow with a person in the loop, and what Nexra keeps as proof.
Top actions
- Create batch jobApproval by rule
- Moderate a conversationApproval by rule
- List agent version aliases (beta)Read
Example flow
- Mistral AI · List agent version aliases (beta)Done
- Mistral AI · Create batch jobNeeds approval
- Nexra · Record the decision and result in the evidence chainProven
Example workflow. Routine runs go through on their own. Only a change your rules flag, such as one over a limit you set, waits for the right person, who sees the exact change and the rule before it reaches Mistral AI.
What gets logged
- Whether it ran on its own or who approved it, and under which rule
- A fingerprint of the exact request sent to Mistral AI
- Mistral AI's reply: what it created or changed
- One entry in the evidence chain, linked to the one before
Try this in the sandbox
Run the whole flow with test data in your sandbox, then connect your own account when you are ready to go live.
For developers
Technical detail
Embeddings, extraction, OCR, moderation, classification, files, batches, models, agents, conversations, libraries, skills, RAG indexes, workflow runs and Admin API usage from Mistral AI.
Mistral AI API docs- Source actions
- 36
- Destination actions
- 9
- Data types
- 31
What you can map
Reads are always available. Writes are off until you switch them on for a connection, and governed writes follow your policies.
Agent aliases1
- List agent version aliases (beta)Source
Aliases (for example production) that point at agent versions.
Agent versions2
- Get agent version (beta)Source
One saved version of an agent (instructions are not returned).
- List agent versions (beta)Source
Saved versions of one agent with model and tool types (instructions are not returned).
Agents2
- Get agent (beta)Source
One agent, optionally at a specific version.
- List agents (beta)Source
Agents in the workspace with model, instructions and tool types.
API keys1
- List API keys (beta)Source
API key metadata (hidden key, workspace, product, expiry, last use).
Audit logs1
- List audit logs (beta)SourceIncremental
Organization audit events, oldest first after a point in time; actor and target metadata are not mapped.
Batch jobs3
- Create batch jobDestinationNot retried automaticallyApproval before production
Starts a discounted batch job for one endpoint from uploaded JSONL files or inline requests.
- Get batch jobSource
Status, request counts and output file for one batch job.
- List batch jobsSourceIncremental
Batch inference jobs, newest first, with request counts and output files.
Chat moderation1
- Moderate a conversationDestinationNot retried automaticallyApproval before production
Classifies a conversation (user and assistant turns) against Mistral's moderation categories, judging the last message in context.
Classifications1
- Classify textDestinationNot retried automaticallyApproval before production
Scores text with a fine-tuned classifier model; returns per-target label scores.
Conversations2
- Get conversation (beta)Source
Metadata for one conversation.
- List conversations (beta)Source
Conversation metadata (model or agent, name, timestamps); no message content.
Embeddings1
- Create embeddingsDestinationNot retried automaticallyApproval before production
Vector embeddings for one text or a list of texts, one record per input in order.
Extraction1
- Extract structured dataDestinationNot retried automaticallyApproval before production
Extracts a JSON object that conforms to your schema using chat completions with a json_schema response format.
Files2
- Get file metadataSource
Metadata for one uploaded file (no contents).
- List filesSource
Uploaded file metadata (no file contents).
Fine tuned models1
- Update fine-tuned modelDestinationProvider duplicate protectionApproval before production
Renames a fine-tuned model or changes its description.
Libraries2
- Get library (beta)Source
One document library.
- List libraries (beta)Source
Document libraries used for retrieval, with document counts and size.
Library document status1
- Get library document status (beta)Source
Processing status of one library document, for polling after upload.
Library documents2
- Get library document (beta)Source
Processing status and metadata for one library document.
- List library documents (beta)Source
Documents in a library with processing status, size and summary.
Models2
- Get modelSource
One base or fine-tuned model with context length, capabilities, aliases and deprecation date.
- List ModelsSource
List operation for Models.
Moderation1
- Classify content for moderationDestinationNot retried automaticallyApproval before production
Classifies text against Mistral's moderation categories (for example pii, financial, hate_and_discrimination).
Observability datasets2
- Get observability datasetSource
One evaluation dataset's metadata.
- List observability datasetsSource
Evaluation datasets in AI Studio observability (records are not read).
Ocr1
- Read document with OCRDestinationNot retried automaticallyApproval before production
Converts a PDF or image at a public HTTPS URL into per-page Markdown with tables, using Mistral Document AI.
Rag indexes2
- Get managed RAG indexSource
One managed retrieval index by name.
- List managed RAG indexesSource
Managed retrieval indexes with status, metadata fields and embedding configuration.
Rate limits1
- Get rate limits (beta)Source
Requests per second and per-model token limits for the organization.
Roles1
- List roles (beta)Source
Organization and workspace roles, including custom roles.
Skill versions1
- List skill versionsSource
Every version of one skill with notes and aliases (no skill body).
Skills2
- Get skillSource
One skill, optionally at a version or alias (no skill body).
- List skillsSource
Skills in the workspace with version, aliases and sharing scope (no skill body).
Spend limits1
- Get spend limit (beta)Source
Monthly spend limit, current usage and payment-failure flags for the organization.
Text1
- Create TextDestinationNot retried automaticallyApproval before production
Create operation for Text.
Usage1
- Get monthly usage (beta)Source
Organization usage and prices for one month by product (chat, completion, OCR, audio, connectors, libraries, fine-tuning).
User groups2
- Get user group (beta)Source
One user group.
- List user groups (beta)Source
User groups with organization role and nesting (members are not read).
Workflow runs2
- Get workflow runSource
Status and timing of one workflow run (the result payload is not returned).
- List workflow runsSourceIncremental
Workflow executions with status, timing and deployment (no inputs or results).
Workspaces1
- List workspaces (beta)Source
Workspaces in the organization with member counts and spend limit.
Access and limits
- Authentication
- Workspace API key (plus optional admin API key)
- Plans and access
- La Plateforme workspaces on the Experiment (free) or Scale plan. Rate limits and model access depend on the workspace tier. Admin API reads need an admin-scoped API key; workspace keys are rejected there.
- Test environment
- The Experiment plan offers free, rate-limited access suitable for testing.
- Rate limits
- Requests per second and tokens per minute/month per workspace tier; HTTP 429.
- Pagination
- Files, batch jobs and conversations use zero-based page numbers; agents and libraries use next_page_token; library documents report has_more; admin API keys use offsets and admin workspaces one-based pages.
- Provider events
- None. Flows run on demand, on a schedule, or when a scheduled check finds new or changed records.
Before you build a flow
- Authenticated model discovery and bounded text generation. Select a model available to your key. Generation sends mapped content to the provider and incurs provider charges. Automatic retries are disabled; inspect usage before repeating an uncertain request. No agents, tool calls, image generation or embeddings.
- Embeddings, extraction, OCR, moderation (text and conversation), classification and batch job creation are billable and never retried automatically.
- OCR reads public HTTPS document or image URLs; uploading files (multipart), base64 documents, file downloads and signed URLs are not implemented.
- Batch job creation accepts up to 500 inline requests per Nexra call, or uploaded JSONL input files.
- Agents, conversations and libraries are beta APIs in Mistral's specification; GET /v1/agents is deprecated, so agents are listed through /v1/agents/pages.
- The Admin API (usage, API keys, workspaces, rate limits and audit logs) is beta in Mistral's specification and needs the optional admin API key. API keys are read only and only the hidden key is mapped; audit actor and target metadata (which can hold user details) is not mapped.
- Read only for administration: creating API keys, user, role and group changes, spend limit changes, and deleting or archiving models, files, agents, libraries or documents are intentionally excluded.
- Audio transcription and speech need multipart uploads or binary responses and are not implemented. Fine-tuned model archiving and all deletes are intentionally excluded.
Developer reference
- Create the Mistral AI credentials described above with the narrowest permissions your flows need.
- In Nexra, open Connect, then Connections, then New connection, choose Mistral AI, enter its settings and credentials, and tick only the write actions you need. Connection guide
- Test the connection, then use its actions as flow sources and destinations. Flow guide
Machine-readable detail: GET /api/public/connect/connectors/mistral and the Connections API.