All upcoming models

Google DeepMind · Shipped

Gemini 3.8 Live release date

Released

Sep 15, 2026

What we know

Speech-to-speech model with real-time audio interaction and the ability to interrupt mid-response, similar to OpenAI's GPT-Live family.

Spec

Params
Not stated
Active
Not stated
Context
Not stated
Weights
Closed
Modality
Not stated
Codename
Not stated

Every claim

8 from 6 sources
  • The Verge AIFrom the labSep 24

    Google released Gemini 3.8 Live with Live Avatar feature for Gemini Enterprise customers.

    no date givenno datesource
  • TechmemeFrom the labSep 24

    Google updates Gemini 3.8 Live with Live Avatar feature for Gemini Enterprise customers.

    points at Sep 24exact datesource
  • Google Models & ResearchFrom the labSep 24

    Google's blog announces Gemini 3.8 Live as introduced on the document date, adding near real-time visual presence to Gemini's conversational AI.

    points at Sep 24exact datesource
  • Google Models & ResearchFrom the labSep 15

    Google announces Gemini 3.8 Live, an advanced live dialogue model for natural conversation.

    no date givenno datesource
  • The DecoderPressSep 15

    Google DeepMind released Gemini 3.8 Live, an audio model for developers priced at $1.38 per hour.

    no date givenno datesource
  • TestingCatalogFrom the labSep 15

    Google is rolling out Gemini 3.8 Live with real-time dialogue and visual grounding capabilities.

    no date givenno datesource
  • TechmemeFrom the labSep 15

    Google launched Gemini 3.8 Live, described as one of its most advanced live dialogue models for voice agents.

    no date givenno datesource
  • Simon WillisonFrom the labSep 15

    Google released Gemini 3.8 Live, a speech-to-speech model similar to OpenAI's GPT-Live family.

    points at Sep 15exact datesource
Official announcement