DCXTribeDCXTribe
Back to library
AI
tutorial

Enable Streaming for Near-Instant Response Feedback

Reduce waiting times by streaming responses in your custom AI apps.

Implement the 'generateContentStream' method in your Gemini API calls to display text to users in real-time as it is generated.

Google Gemini

The Scenario

You are building a custom internal tool for your team to summarize long meeting transcripts and want the interface to feel fast and responsive.

Before & after

The old way

Manually Waiting for a full text generation often results in a 'loading' state where the user sees nothing until the entire paragraph is complete, taking 15-30 seconds.

With AI

Using the Gemini API with streaming, chunks of the response appear as they are generated, letting the user start reading immediately. This reduces perceived latency to about 2-3 seconds for the first word.

The Prompt

You are a developer using Node.js. Implement a streaming response using the Gemini SDK: 

const { GoogleGenerativeAI } = require("@google/generative-ai");
const genAI = new GoogleGenerativeAI([YOUR_API_KEY]);
const model = genAI.getGenerativeModel({ model: "gemini-pro"});

async function run() {
  const prompt = "[DESCRIBE_YOUR_TASK_HERE]";
  const result = await model.generateContentStream(prompt);
  for await (const chunk of result.stream) {
    const chunkText = chunk.text();
    process.stdout.write(chunkText);
  }
}
run();

Streaming is particularly useful for longer creative writing or technical explanations where the user can begin consuming the output while the rest is being calculated.

Source

Release notes  |  Gemini API  |  Google AI for Developers
"we'll use streaming for faster interactions we'll also take a look"