Watch how we used 3.8 Live Extended Thinking to act as a programming tutor. Both models feature: 🔵 Upgraded reasoning 🔵 Near real-time visual understanding 🔵 Automatic detection for 97 languages 🔵 Background tool calling without disrupting your chat For your most difficult tasks, 3.8 Live Extended Thinking adds... increased performance and precision – narrating task progress to keep the conversation going. Try it now in Gemini Live in the Google Gemini or start building with the Gemini API via Google AI Studio. Find out more →show more

We’re introducing Gemini 3.8 Live and 3.8 Live Extended Thinking – our best conversational AI. The models talk, think, and handle tasks in the background without breaking your flow. 🧵

holy shit why does it talk like that

love this, hate the voice tbh

all nice but, couldn't you pick a more annoying voice? 🤣💀

Smart, but did you guys train this voice on voice actors for American commercials?

This looks really cool, but why does that voice sound like it's had about three too many Monster energy drinks?

thats the worst voice you guys could've picked! well done.

tbf a tutor that doesn't lose its rag when you ask the same question five times is genuinely useful. bit nicer than my old CS teacher, not gonna lie

why does it talk so funny? haha

I hope not a Chinese replying AI this time for Turkish speech

Do we have this on Windows as well ? On the desktop app ?

this voice sucks thooo

This is where AI tutors get really interesting. Seeing what you’re working on is useful, but being able to show the explanation back matters too. That’s the direction I’m taking @OpenLume: ask a technical question, get a visual explainer video, then interrupt it with follow-ups when something doesn’t click.

I tried it. It is very dumb. It claimed Astra is a Google model, that OpenAI have o1 and o3 only. I won't be using it. Thank you.

to be honest, I prefer a robotic voice instead

who the fuck picked this goofy ass voice

Sound like goofy from Mickey Mouse

Is one allowed to commit violence on a computer for an insultingly peppy voice?

The future of programming tutoring is a model that can explain your bug while silently judging the variable you named temp_final_final.

Best voice model in my opinion. Chat Live is 2nd but Gemini’s new model is so good and connects to Google Drive / Workspace on the fly. Great job Google Team!

把 extended thinking 直接接到真实 IDE 工作流，价值不只是更长的推理，而是让模型在复杂任务里保持上下文连续。下一步关键会是：如何把推理预算和可验证的代码变更绑定起来。

Reproducible probes and transport-level traces make these model comparisons so much more useful.

the bar is a tutor that stays on the broken line. no alt-tab to explain the bug.

Wait, it tells you to write the code yourself? That’s so 2025

말로 가르치면서도 흐름이 안 끊기는 쪽이, 실제 튜터에 가깝네요.

@heyclicky u can now make a wispr flow “google is coming for us” video

Talking and thinking in the same breath now.

Programming tutor mode, now with 97 languages of homework.

GSD on a whole new level

Tragam as vozes do gemini live do AI studio para o gemini App, estamos cansados daquelas vozes antigas

picked the worst domain for this demo. literally nobody codes anymore and why would anyone use this to programming 😭 so many better use cases like photoshop/illustrator/premier/blender

I had been sitting with this idea since past month and started working on it but i guess i am late on this one too. Anyways, congratulations to the deepmind team for this. I hope this expands as an educational milestone for students all over the world in the generation of agentic era.

This looks so annoying 😂

Just fyi, y’all are going backwards