Sends a moderation request for a chat conversation to the Mistral API.

Accepts either a JSON string or an array of chat messages (Messages) representing the conversation, and returns a ModerationResponse with flagged content and category breakdowns.

Arguments:

• model : String — the name of the moderation model to use (e.g., "mistral-moderation-latest")
• input : Array(Messages) | String — the conversation history either as a parsed array or a JSON string

Returns:

• ModerationResponse — the parsed moderation result for the chat conversation

Raises:

• ArgumentError if either model or input is empty
• RuntimeError via handle_error if the response indicates an error

Example:

# Use custom Messages type
messages = [
  Messages.new(role: "user", content: "I want to hate someone."),
  Messages.new(role: "assistant", content: "That doesn't sound good."),
]

# or String
messages = %([{"role": "user", "content": "I want to hate someone."}])

# if you just pass the string
messages = “AAA WHAT TIME IS IT?!”
# string convert to -> [Messages.new(role: "user", content: messages)]

response = client.chat_moderations(
  model: "mistral-moderation-latest",
  input: messages
)

puts "Content: #{response.results[0].categories}"

Sends a classification request to the Mistral API (/v1/classifications).

This method takes a plain input string and sends it to the classification endpoint using the specified model. It returns the raw response body as a string, which can be further parsed depending on the expected response structure.

Arguments:

• model : String — the name of the classification model to use (e.g., "mistral-classification-latest")
• input : String — the text to classify

Returns:

• String — raw JSON response body from the API

Raises:

• ArgumentError if model or input is empty
• RuntimeError via handle_error if the response indicates an error

Example:

response_json = client.classification(
  model: "mistral-classification-latest",
  input: "This document discusses violent actions."
)

puts response_json

TODO learn how the classifier works properly and create types

Sends a classification request for a chat conversation to the Mistral API (/v1/chat/classifications).

This method accepts either:

• an array of chat messages (Array(Messages)), or
• a JSON string (representing an array of messages), or
• a plain string (treated as a single user message)

The input is normalized to a list of Messages, wrapped in a ClassifiersChatRequest, and sent to the API. Returns the raw JSON response from the server.

Arguments:

• model : String — the name of the classification model (e.g., "mistral-chat-classification-latest")
• input : Array(Messages) | String — conversation history as array or JSON or plain string

Returns:

• String — raw JSON response body from the API

Raises:

• ArgumentError if model is empty or input type is unsupported
• RuntimeError via handle_error if the response status is an error

messages = [
  Messages.new(role: "user", content: "I hate everything."),
  Messages.new(role: "assistant", content: "Let's keep things peaceful."),
]
# if you just pass the string
messages = “AAA WHAT TIME IS IT?!”
# string convert to -> [Messages.new(role: "user", content: messages)]

response = client.classification_chat(
  model: "mistral-chat-classification-latest",
  input: messages
)

puts response

TODO learn how the classifier works properly and create types

Sends a moderation request to the Mistral.

This method checks the input text against moderation classifiers to determine whether the content violates usage policies.

Arguments:

• model : String — the name of the moderation model to use (e.g., "mistral-moderation-latest")
• input : String — the text to moderate

Returns:

• ModerationResponse — the parsed moderation result from the API

Raises:

• ArgumentError if either model or input is empty
• RuntimeError via handle_error if the response indicates an error

Example:

response = client.moderation(
  model: "mistral-moderation-latest",
  input: "I hate you and want to hurt people"
)

puts "Content: #{response.results[0].categories}"