Google Gemini - Send messages to Gemini models action

The Send messages to Gemini models action sends a message to and receives a response from a Gemini model you specify. Fields are loaded dynamically based on the Message type you select. For example, the following are the input and output schemas if you select Single message as the Message type:

Input

Input field	Description
Model	Select the Gemini model to use.
Message type	Select the type of message to send.
Text to send	Enter a message to send to Gemini.
Category	Select a safety category to evaluate, such as HARM_CATEGORY_HATE_SPEECH.
Threshold	Select the value threshold that a message must meet in the selected Category in order to block it.
Stop sequence	Provide a list of strings that cause the model to stop generating text.
Temperature	Enter a number to control the randomness of the model's output. A higher Temperature results in a more random output, while a lower Temperature results in a more predictable output.
Max output tokens	Specify the maximum number of tokens that the model can generate.
TopP	Specify a number to control the probability of the model generating each token. A higher TopP results in the model generating more likely tokens, while a lower TopP results in the model generating more unlikely tokens. Allowed values include any decimal value between 0 and 1.
TopK	Specify a number to control the number of tokens that the model considers when generating each token. A higher TopK results in the model considering more tokens, while a lower TopK results in the model considering fewer tokens. Allowed values include any positive integer.

Output

Output field	Description
Gemini reply	The response from Gemini.
Sexually explicit	The likelihood that the text contains sexually explicit content. NEGLIGIBLE indicates there is little to no risk of such content being present.
Hate speech	The likelihood that the text contains or promotes hate speech, such as discriminatory symbols or context targeting protected groups. NEGLIGIBLE indicates there is little to no risk of such content being present.
Harassment	The likelihood that the text contains harassing behavior, threats, or abuse toward individuals or groups. NEGLIGIBLE indicates there is little to no risk of such content being present.
Dangerous content	The likelihood that the text contains dangerous or harmful content such as violence, self-harm, or instructions for unsafe behavior. NEGLIGIBLE indicates there is little to no risk of such content being present.