fix api key inconsistency and cleanup in live api quickstart

[knQzx]

Summary

• fixed undefined GOOGLE_API_KEY variable — now reads from os.environ like the rest of the notebook
• added a warning when native-audio model is selected since it doesn't support TEXT response modality
• removed redundant client/model redefinition in the resumable sessions section

Test plan

• verified notebook cells reference the correct api key variable
• verified model guard prints warning for native-audio models
• verified resumable sessions section works without redundant client init

fixes #948

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.

[gemini-code-assist]

Note

Gemini is unable to generate a summary for this pull request due to the file types involved not being currently supported.

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[knQzx]

I have signed the CLA

[Venkaiahbabuneelam]

Hi @knQzx, Thanks for reaching out us!

Regarding you PR few checks were failing, Could you please have a look once.

Thanks

[knQzx]

pushed nbfmt formatting only - content is identical just whitespace so notebook format should pass

notebook lint will still fail on google::inclusive_language - it flags the word native in the model id gemini-2.5-flash-native-audio-preview-09-2025 which is a product name and can't be renamed

same thing happens on main for this file - lint just doesn't surface there because it only runs on notebooks touched by a pr - 5 other notebooks in main also reference native-audio (Browser_as_a_tool, LiveAPI_plotting_and_mapping, Get_started_LiveAPI_tools, Models, Search_Grounding)

the native-audio guard is literally the fix for #948 so removing it to please the linter doesn't make sense

on your side you could add --exclude_lint=google::inclusive_language to the workflow or whitelist product ids in the wordlist

also the run is stuck in action_required - could you approve it

[github-actions]

Marking this pull request as stale since it has been open for 14 days with no activity. This PR will be closed if no further activity occurs.

[kkorpal]

When I run this cell in Colab, it throws a NameError: name 'GOOGLE_API_KEY' is not defined. We need to fetch the key from the environment.

[thorwebdev]

@knQzx this is great, thanks so much. Could you also update the send_client_content to use await session.send_realtime_input(text="Hello, how are you?") instead and use gemini-3.1-flash-live-preview as the default? https://ai.google.dev/gemini-api/docs/live-api/capabilities#send-text

[knQzx]

@thorwebdev done in 161c951

• replaced send_client_content with send_realtime_input(text=...) in all 4 call sites (Text to Text, Simple text to audio, AudioLoop.send, resumable sessions send)
• default MODEL bumped to gemini-3.1-flash-live-preview, kept the other ids in the dropdown so users can still pick them

native-audio guard left in place - now only fires if someone manually picks the native-audio model from the dropdown

[thorwebdev]

Hey @knQzx I'm sorry, I should have mentioned that gemini-3.1-flash-live-preview is also a native audio model and also does not support "response_modalities": ["TEXT"], only AUDIO, so this would need a bit of a larger refactor.

Would you be up for changing every cell to AUDIO by any chance and removing the initial Text to Text cell, seeing that the only model that used to support this has since been deprecated? https://ai.google.dev/gemini-api/docs/deprecations#live-api-models

Sorry about that!

[thorwebdev]

@knQzx I've made the update here now: #1222

I really appreciate you getting started on these changes and contributing your time!

[knQzx]

@knQzx I've made the update here now: #1222

I really appreciate you getting started on these changes and contributing your time!

thanks @thorwebdev, all good. closin this in favor of #1222 which has the broader cleanup anyway. happy to help on the next one!!