Google announced the launch of Gemini 3.5 Live Translate on June 9 on its official blog. This is the latest audio model for the Gemini Live API. Its goal is to eliminate conversation pauses caused by waiting for a whole sentence to finish in real-time speech translation. Gemini 3.5 Live Translate automatically detects 70+ languages and outputs translations using a continuous generation approach.
Confirmed technical features of Gemini 3.5 Live Translate
Translation mechanism: Gemini 3.5 Live Translate uses a “speech-to-speech” translation approach. It dynamically adjusts in real time between “waiting for more context to improve accuracy” and “outputting immediately to keep up with the speaker,” with the overall translation gap of a few seconds.
Speech feature retention: The translated output preserves the speaker’s intonation, pacing, and pitch, rather than presenting everything in a uniform machine voice.
Differences from its predecessor (confirmed by Google): Previously, Google’s speech translation system had to wait for the speaker to finish a complete sentence before starting translation, causing interruptions during conversations; Gemini 3.5 Live Translate replaces this waiting mechanism with continuous generation.
Three confirmed launch channels on the same day
Developer public preview: Gemini Live API and Google AI Studio will open for access starting June 9, 2026.
Enterprise private preview: Google Meet for Enterprise will begin a private preview starting in June 2026.
Consumer global update: The Google Translate app for Android and iOS has been updated in parallel. Android also adds a new “Listening Mode.” When users hold the phone close to their ear, translated speech plays from the earpiece—without needing headphones and without affecting the surrounding environment. Google cites example use cases including foreign-language tours outside museums or receiving foreign-language phone calls in quiet settings.
Confirmed competitive landscape and partners
Competing products in the same category (already launched): Meta SeamlessM4T, Samsung Galaxy AI real-time call translation, Apple Live Translation, OpenAI Realtime API.
Grab integration testing (confirmed): The Southeast Asia ride-hailing platform Grab is testing Gemini 3.5 Live Translate to enable real-time multilingual communication between drivers and passengers. Grab users make more than ten million voice calls per month through its platform, with languages including Thai, Vietnamese, Malay, Indonesian, and Filipino.
Feedback from early partners (confirmed): CJ ENM and LiveKit say translation quality, accuracy, and latency all meet expectations.
FAQ
What are the main differences between Gemini 3.5 Live Translate and Google’s earlier speech translation features?
According to Google’s official explanation, the main difference lies in the timing of when translation happens. Previously, speech translation had to wait for the speaker to finish a complete sentence before starting translation, which interrupted the conversation; Gemini 3.5 Live Translate uses a continuous generation approach, reducing the overall gap to a few seconds, while also preserving the speaker’s intonation, pacing, and pitch.
How do you use Android “Listening Mode,” and what are the applicable scenarios?
“Listening Mode” is a new feature in the Android version of Google Translate. When users hold the phone close to their ear, translated speech plays from the earpiece. It does not require pairing with Bluetooth headphones and will not affect the surrounding environment through external speakers. Google lists applicable scenarios including foreign-language tours in museums and receiving foreign-language calls in quiet settings.
How can I get Gemini 3.5 Live Translate right now?
As of June 9, 2026, Gemini 3.5 Live Translate is available through three channels: the developer public preview of the Gemini Live API and Google AI Studio; the enterprise private preview of Google Meet (to begin this month); and the global update of the Google Translate apps for Android and iOS.
