Translating Text with Gemini AI

Gemini AI provides translation capabilities using the text generation models.

Model Selection

Choose between cost and quality:

• Gemini 3 Flash (gemini-3-flash-preview): Cost-efficient

  • $0.50 input / $3.00 output per 1M tokens
  • 40x cheaper than Pro

• Gemini 3 Pro (gemini-3-pro-preview): Premium quality

  • $2.00 input / $120.00 output per 1M tokens

Single Translation

import { GoogleGenAI } from '@google/genai';

const genAI = new GoogleGenAI(process.env.GEMINI_API_KEY);
const model = genAI.getGenerativeModel({ model: 'gemini-3-flash-preview' });

const prompt = `Translate the following Japanese text to English:\n\n${text}`;

const response = await model.generateContent(prompt);
const translatedText = response.response.text();

Batch Translation

More efficient for multiple items:

const prompt = `Translate the following Japanese titles to English.
Return JSON format: [{"original": "...", "translated": "..."}]

Titles:
${titles.map((t, i) => `${i + 1}. ${t}`).join('\n')}
`;

const response = await model.generateContent(prompt);
const result = JSON.parse(response.response.text());

Response Parsing

Gemini wraps JSON in markdown code blocks:

function parseGeminiJSON(text: string): any {
  // Remove markdown code blocks
  const cleaned = text.replace(/```(?:json)?\s*([\s\S]*?)\s*```/, '$1');
  return JSON.parse(cleaned);
}

Configuration

const response = await genAI.models.generateContent({
  model: 'gemini-3-flash-preview',
  contents: prompt,
  config: {
    generationConfig: {
      temperature: 0.7,
      maxOutputTokens: 1000,
    },
  },
});