Limits and specifications of the Live API

This page describes various limits and specifications for using the Live API and its models.

Session-related limits

For the Live API , a session refers to a persistent connection where input and output are streamed continuously over the same connection.

If the session exceeds any of the following limits, the connection is terminated.

• Connection lengthis limited to around 10 minutes.

• Session lengthdepends on the input modalities:

  • Audio-only input sessions are limited to 15 minutes.
  • Video + audio input are limited to 2 minutes.

• Session context windowis limited to 128k tokens.

Rate limits

The Live API has rate limits for both concurrent sessions per Firebase project as well as tokens per minute (TPM).

• Gemini Developer API :

  • Limits vary based on your project's Gemini Developer API "usage tier" (see their rate limits documentation )

• Vertex AI Gemini API :

  • 1,000 concurrent sessions per Firebase project
  • 4M tokens per minute

Audio formats

The Live API supports the following audio formats:

• Input audio format:Raw 16 bit PCM audio at 16kHz little-endian

• Output audio format:Raw 16 bit PCM audio at 24kHz little-endian

• Supported MIME types: audio/x-aac , audio/flac , audio/mp3 , audio/m4a , audio/mpeg , audio/mpga , audio/mp4 , audio/ogg , audio/pcm , audio/wav , audio/webm

To convey the sample rate of input audio, set the MIME type of each audio-containing Blob to a value like audio/pcm;rate=16000 .

Video formats

The Live API expects a sequence of discrete image frames and supports video frames input at 1 frame per second (FPS).

• Recommended input: native 768x768 resolution at 1 FPS.

• Supported MIME types: video/x-flv , video/quicktime , video/mpeg , video/mpegs , video/mpg , video/mp4 , video/webm , video/wmv , video/3gpp

Note that this specification makes the Live API unsuitable for use cases that require analyzing fast-changing video, such as play-by-play in high-speed sports.

Response voices

The Live API supports the following response voice options. For demos of what each voice sounds like, see Chirp 3: HD voices .

If you don't specify a response voice, the default is Puck .

Learn how to specify the response voice .

Languages

The Live API supports the following languages. Learn how to influence the response language .