gemini-2.5-flash-native-audio-latest · Live API

Talk to AI, Naturally.

Real-time voice conversation powered by Google Gemini's native audio model. No scripts, no delays — just natural, flowing dialogue.

1-minute free trial · No sign-up needed · Works in Chrome & Edge

What makes it different

< 300 ms

Full-Duplex Audio

Bidirectional streaming under 300 ms. You and the AI can talk simultaneously — real interruption support, no push-to-talk.

Gemini 2.5 Flash

Gemini Native Voice

Powered by Google's gemini-2.5-flash native audio model. Natural prosody, expressive tone, human-like rhythm.

Real-time

Live Transcription

Both sides of the conversation are transcribed in real-time — read what you said and what the AI replied, all as it happens.

How it works

01

Click Start Session

Opens a secure WebSocket connection to the AI backend.

02

AI Greets You First

The assistant speaks an opening welcome automatically.

03

Click the Mic Button

Your browser streams audio; Gemini VAD detects speech.

04

AI Responds Live

Audio streams back and plays instantly — no waiting.

Ready to have a conversation?

Open the demo, click Start Session, then the mic button — and just talk.

Open the Demo