Google Translate Gets Major Gemini Boost

On December 12, 2025, Google announced that it had rolled out a major upgrade to Google Translate, powered by its latest Gemini models.

The update brings together three key developments: improved AI translation quality, a new AI live speech translation experience (currently in beta), and an expansion of supported languages for practice and skill-building features within the Google Translate app.

The core update integrates Google’s newest Gemini model into Google Translate. According to Google, the model is designed to better handle idioms, slang, and conversational language.

“Starting today, Google Translate uses advanced Gemini capabilities to better improve translations on phrases with more nuanced meanings like idioms, local expressions or slang,” the company said. 

The update is rolling out initially in the U.S. and India, supporting English and nearly 20 other languages — including Spanish, German, Hindi, Japanese, and Chinese, with additional languages planned — and is available across Android, iOS, and the web.

AI Live Speech Translation Now in Headphones

Google also announced an upgrade to its AI live speech translation capabilities.

In late August 2025, Google introduced AI live speech translation in Google Translate, enabling users to speak into their device, hear translations aloud, and see text on screen.

Now, Google Translate uses Gemini’s native speech-to-speech model, allowing users to hear real-time translations directly through their headphones. “Building on Gemini’s new live speech-to-speech translation capabilities, we are rolling out a beta experience enabling you to hear real-time translations in your headphones,” the company said.

The new system supports two distinct modes:

• Continuous listening — where Gemini automatically listens to speech in multiple languages and translates it into a single target language, allowing users to hear their surroundings in real time.
• Two-way conversation — where Gemini handles translation between two languages in real time, automatically switching the output language based on who is speaking.

According to Google, this enables a range of use cases beyond one-to-one conversations. “Whether you’re trying to have a conversation in a different language, listen to a speech or lecture while abroad, or watch a TV show or film in another language, you can now put in your headphones, open the Translate app, tap “Live translate” and hear a real-time translation in your preferred language,” the company said.

The system supports more than 70 languages and 2,000 language pairs and delivers more natural-sounding speech by preserving tone and speaking style. In addition to multilingual input, which allows the system to understand multiple languages within a single session, additional capabilities include automatic language detection, removing the need for manual language selection, and noise robustness, enabling translation in loud or outdoor environments.

The company said the feature is currently available in beta in the Translate app on Android in the US, Mexico, and India, works with any pair of headphones, and is expected to roll out more broadly in 2026 alongside iOS support.

Google’s move follows similar efforts by other big tech companies, including Apple, which announced earlier this year on-device AI live speech translation.

Last but not least, Google is also expanding its language learning tools with improved feedback, giving users tips based on their speaking practice and making it easier to track learning progress over time, and extending the features to additional countries and language pairs.