Skip to content

docs(realtime): describe the lazy Gemini Live setup built from the first session.update - #1216

Open
mateo-berri wants to merge 2 commits into
mainfrom
litellm_gemini_live_lazy_setup_docs
Open

docs(realtime): describe the lazy Gemini Live setup built from the first session.update#1216
mateo-berri wants to merge 2 commits into
mainfrom
litellm_gemini_live_lazy_setup_docs

Conversation

@mateo-berri

@mateo-berri mateo-berri commented Sep 5, 2026

Copy link
Copy Markdown
Contributor

TLDR

  • Gemini and Vertex Live setup is now built from the client's first session.update, so output_modalities and modalities apply
  • Documents the connect-time session.created, the session.updated acknowledgement, and the text-to-audio downgrade on audio-only models
  • Adds the LITELLM_GEMINI_LIVE_DEFER_SETUP row to the config settings table

Pairs with BerriAI/litellm#39864 (LIT-6974)

Merge after BerriAI/litellm#39864 lands, since these pages describe behavior that only exists once that PR ships

Type

📖 Documentation


Note

Low Risk
Documentation-only changes with no runtime impact; behavior described depends on the linked implementation PR landing first.

Overview
Updates Gemini and Vertex realtime provider docs and the LITELLM_GEMINI_LIVE_DEFER_SETUP env reference to match lazy Live setup behavior (companion implementation PR).

session.update is no longer described as ignored or connect-time-only: every update before the first content frame (conversation item, audio, or tool result) is merged into the single Gemini/Vertex setup, with session.updated acknowledgement; updates after content are ignored because setup is one-shot per connection. Default path sends setup with the first content frame, using merged session.update fields or model defaults when none were sent.

gemini_live_defer_setup is reframed as optional (default false): enable it only when clients stream audio or text before their session.update, so LiteLLM buffers that early traffic until the update can still shape setup. Vertex docs also note output_modalities: ["text"] on audio-only models is downgraded to audio with a proxy warning.

Reviewed by Cursor Bugbot for commit 6a8e189. Bugbot is set up for automated code reviews on this repo. Configure here.

@vercel

vercel Bot commented Sep 5, 2026

Copy link
Copy Markdown

The latest updates on your projects. Learn more about Vercel for GitHub.

Project Deployment Actions Updated
litellm Ready Ready Preview Sep 5, 2026 6:00am UTC

Request Review

@mateo-berri

Copy link
Copy Markdown
Contributor Author

bugbot run

@cursor cursor Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Stale Bugbot comment from a previous run.

@mateo-berri

Copy link
Copy Markdown
Contributor Author

bugbot run

@cursor cursor Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

Reviewed by Cursor Bugbot for commit 6a8e189. Configure here.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant