Skip to content

docs(live): restore Gemini 3.1 Flash Live and fix broken anchors - #2208

Merged
joefernandez merged 3 commits into
mainfrom
kaz-live-models-fix
Sep 8, 2026
Merged

joefernandez merged 3 commits into
mainfrom
kaz-live-models-fix

Conversation

@kazunori279

Copy link
Copy Markdown
Collaborator

Follow-up to #2086, reviewing the models content and the redirect/link changes as requested.

Limits and quotas: no change needed

I checked the new Platform limits and quotas table against upstream and it is right, so it stays as merged:

Limit Verified
Session duration, audio-only — 15 min (both backends) ✅
Session duration, audio + video — 2 min (both backends) ✅
Connection lifetime — ~10 min ✅
Agent Platform conversation session default — 10 min ✅
Concurrent sessions — 1,000/project pay-as-you-go, unlimited with Provisioned Throughput ✅

My earlier numbers were the wrong ones: I had Agent Platform capping every session at 10 minutes, which conflated the connection limit with the session limit.

Models: one real regression

gemini-3.1-flash-live-preview was dropped from the page, but the model exists — released 2026-03-26, documented on AI Studio, and listed alongside 2.5 in the Live API capabilities guide. Agent Platform supports no Gemini Live 3.x model, which is what makes the AI-Studio-only distinction worth stating rather than omitting.

This PR restores it inside the merged structure rather than reverting:

  • Two model rows with launch stages. gemini-live-2.5-flash-native-audio is the only GA Live model; every AI Studio ID is preview.
  • The per-model feature table gets a second column. With one column it conveyed nothing; with two it answers the question a reader actually has. 3.1 supports neither proactivity/affective dialog nor non-blocking tools, and uses thinking_level instead of thinking_budget.
  • The global location note goes back to stating the restriction as fact — Live 2.5 models are not served at GOOGLE_CLOUD_LOCATION=global — instead of asking the reader to go verify.

Link audit

Checked docs/**/*.md for references to streaming/dev-guide/* and validated the redirect map:

  • Zero references to streaming/dev-guide/* or live/dev-guide/* remain in docs/.
  • All 107 entries in redirect_maps resolve: no source shadows a real page, every target file exists, and all 107 generated redirect stubs in site/ point at a page that was built.
  • No links remain to the removed live/streaming-tools.md or live/voice.md.

Two broken anchors did turn up, both fixed here:

  • live/configuration.md linked to #response-modalities; the heading is ## Response modes.
  • live/evaluation.md linked to #audio-user-simulation-live-agents; the heading is ## Audio user simulation for live agents.

Plus one stale model ID outside the live section: tutorials/multi-tool-agent.md suggested gemini-2.0-flash-live-001, shut down 2025-12-09.

mkdocs build --strict is clean, with no remaining link or anchor warnings.

Verify the models and limits content merged in #2086 against upstream
documentation.

The platform limits table is correct as merged and stays as is: audio-only
sessions cap at 15 minutes and audio+video at 2 minutes on both backends, a
connection lasts ~10 minutes, and Agent Platform additionally defaults a
conversation session to 10 minutes. The earlier claim that Agent Platform
capped every session at 10 minutes conflated the connection limit with the
session limit.

Restore the coverage that was dropped:

- `gemini-3.1-flash-live-preview` is a real model, released 2026-03-26 and
  documented on AI Studio. It is not available on Agent Platform, which
  supports no Gemini Live 3.x model.
- Mark launch stages. `gemini-live-2.5-flash-native-audio` is the only GA Live
  model; the AI Studio IDs are all preview.
- Give the per-model feature table a second column, since with one column it
  said nothing. 3.1 supports neither proactivity/affective dialog nor
  non-blocking tools, and configures thinking with `thinking_level` rather
  than `thinking_budget`.
- State the `global` location restriction as fact rather than as something to
  go check: Live 2.5 models are not served there.

Also fix two anchors that do not resolve, and a deprecated model ID:

- `live/configuration.md` linked to `#response-modalities`; the heading is
  `## Response modes`.
- `live/evaluation.md` linked to `#audio-user-simulation-live-agents`; the
  heading is `## Audio user simulation for live agents`.
- `tutorials/multi-tool-agent.md` suggested `gemini-2.0-flash-live-001`, which
  was shut down on 2025-12-09.
@netlify

netlify Bot commented Sep 7, 2026 •

Copy link
Copy Markdown

✅ Deploy Preview for adk-docs-preview ready!

Name Link
🔨 Latest commit c994dba
🔍 Latest deploy log https://app.netlify.com/projects/adk-docs-preview/deploys/6aa050301a000b00080c0c29
😎 Deploy Preview https://deploy-preview-2208--adk-docs-preview.netlify.app
📱 Preview on mobile
Toggle QR Code...

QR Code

Use your smartphone camera to open QR code link.
🤖 Make changes Run an agent on this branch

To edit notification comments on pull requests, go to your Netlify project configuration.

Comment thread docs/live/models.md Outdated
@joefernandez
joefernandez enabled auto-merge (squash) September 8, 2026 18:13
@joefernandez
joefernandez merged commit 77ac936 into main Sep 8, 2026
11 checks passed
@joefernandez
joefernandez deleted the kaz-live-models-fix branch September 8, 2026 18:13
zyantw pushed a commit to zyantw/adk-docs that referenced this pull request Sep 24, 2026
…gle#2208)

* docs(live): restore Gemini 3.1 Flash Live and fix broken anchors

Verify the models and limits content merged in google#2086 against upstream
documentation.

The platform limits table is correct as merged and stays as is: audio-only
sessions cap at 15 minutes and audio+video at 2 minutes on both backends, a
connection lasts ~10 minutes, and Agent Platform additionally defaults a
conversation session to 10 minutes. The earlier claim that Agent Platform
capped every session at 10 minutes conflated the connection limit with the
session limit.

Restore the coverage that was dropped:

- `gemini-3.1-flash-live-preview` is a real model, released 2026-03-26 and
  documented on AI Studio. It is not available on Agent Platform, which
  supports no Gemini Live 3.x model.
- Mark launch stages. `gemini-live-2.5-flash-native-audio` is the only GA Live
  model; the AI Studio IDs are all preview.
- Give the per-model feature table a second column, since with one column it
  said nothing. 3.1 supports neither proactivity/affective dialog nor
  non-blocking tools, and configures thinking with `thinking_level` rather
  than `thinking_budget`.
- State the `global` location restriction as fact rather than as something to
  go check: Live 2.5 models are not served there.

Also fix two anchors that do not resolve, and a deprecated model ID:

- `live/configuration.md` linked to `#response-modalities`; the heading is
  `## Response modes`.
- `live/evaluation.md` linked to `#audio-user-simulation-live-agents`; the
  heading is `## Audio user simulation for live agents`.
- `tutorials/multi-tool-agent.md` suggested `gemini-2.0-flash-live-001`, which
  was shut down on 2025-12-09.

* Apply suggestion from @joefernandez

---------

Co-authored-by: Joe Fernandez <931947+joefernandez@users.noreply.github.com>
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.

2 participants