gemini-3

Gemini 3 Node.js Example

This example sends a direct Gemini 3 chat request from Node.js through APIDot.

What this example shows

• Loading APIDOT_API_KEY from the environment or the repo-root .env file.
• Calling the Gemini Native endpoint with the model ID in the URL path.
• Handling non-2xx and API-level errors explicitly.
• Printing the response JSON.

Requirements

• Node.js 18 or newer.
• An APIDot API key stored server-side.
• No npm dependencies are required.

Environment variables

Use placeholders only. Do not commit real credentials.

APIDOT_API_KEY=YOUR_API_KEY_HERE

How to run

cd node/gemini-3
cp ../../.env.example ../../.env
# Edit ../../.env and set APIDOT_API_KEY
npm start

Expected response

Shortened response:

{
  "candidates": [
    {
      "content": {
        "role": "model",
        "parts": [
          {
            "text": "Here is a concise launch checklist..."
          }
        ]
      },
      "finishReason": "STOP"
    }
  ]
}

Production notes

• Keep APIDot API keys in server-side environment variables.
• Put the Gemini model ID in the URL path.
• Set request timeouts in your backend HTTP client.
• Log model names and application context, but never log API keys.
• Validate user input before sending it to the model.

Common mistakes

• Committing a real API key or .env file.
• Sending APIDot API keys from browser code.
• Putting the Gemini model ID in the body instead of the URL path.
• Adding async task_id polling to this direct chat request.
• Assuming every response contains at least one candidate.

Related links

• cURL quickstart: ../../curl/chat/gemini-3.md
• Gemini 3 docs: https://apidot.ai/docs/gemini-3
• Chat models: https://apidot.ai/models/chat
• Gemini 3 landing page: https://apidot.ai/models/gemini-3