Wildbots Gemini icon

Wildbots Gemini

Google Gemini AI via Cloudflare proxy — works in any region

Actions15

Text → Message a Model

AI-generated

Summary

Send a list of message exchanges to a specified Google Gemini AI model, optionally using built-in tools for enhanced context, and receive generated text responses from the model.

Inputs

  • Model (required) — Specify the Google Gemini model to message, either by selecting from a searchable list or entering the model ID directly.
  • Messages (required) — Provide one or more message pairs, each including the content text and the role ('user' to send a user message or 'model' to set model tone/personality) defining how the model interprets the message.
  • Simplify Output — Choose whether to return a simplified array of candidate responses or the full raw API response.
  • Output Content as JSON — Select to instruct the model to return its response formatted as JSON data.
  • Built-in Tools — Optionally enable built-in tools such as Code Execution, File Search (with store names and metadata filters), Google Maps (with latitude and longitude), Google Search, and URL Context for dynamic real-time information and content analysis.
  • Options — Configure model generation parameters including frequency and presence penalties, maximum tokens output, number of completions to generate, temperature, top K and top P sampling options, system message overrides, token thinking budget, inclusion of merged response text, and maximum tool call iterations.

Output shape

a list of response objects, each representing a candidate completion generated by the model

When Simplify Output is enabled (default), each item contains a single candidate's response in a simplified structure. Otherwise, the full API response is output including all candidates. If Output Content as JSON is set, the model will attempt to return JSON-formatted responses. Support for built-in tools and external connected tools influences how the model generates responses and may result in iterative tool calls before final completion.

Examples

Example 1: Sending a single message to a Gemini model with default options

Model selected from list, one user message saying 'Hello, how can you help me?', Simplify Output enabled, no built-in tools.

Example 2: Sending multiple messages including model instructions and enabling Google Search

Model ID specified directly, multiple messages with roles 'model' (set behavior) and 'user' (query), built-in tool Google Search enabled to allow real-time web info, and custom temperature set for response randomness.

Links

Discussion