Loading video...

Video Failed to Load

Go Home

the computer use tool in the gemini api is now generally available no breaking changes from preview, just production-ready stability for your browser, mobile, and desktop agents docs here:

305,798 views • 4 days ago •via X (Twitter)

28 Comments

Logan Kilpatrick's profile picture
Logan Kilpatrick4 days ago

next up, hosted computer use!

Bee's profile picture
Bee4 days ago

Now bring it in Antigravity :)

Stats Wire's profile picture
Stats Wire4 days ago

google is working 24/7 these days.

王一's profile picture
王一4 days ago

Computer-use for @antigravity

Veera Tech info's profile picture
Veera Tech info4 days ago

Gemini's computer use is now production ready build browser, mobile, and desktop agents with confidence!

Pedro G's profile picture
Pedro G4 days ago

Generally available so now when it clicks "Delete" instead of "Download" it's with production-ready stability haha

Cipher0k's profile picture
Cipher0k4 days ago

what happens when the page it's clicking through has instructions in white text?

AlleyBoss's profile picture
AlleyBoss4 days ago

whats the point of letting agent control your computer and clicked through frontend ui, its just faster using your hands

Michael Waitze's profile picture
Michael Waitze4 days ago

We actually went deeper on the Gemini API computer use tool and its availability here:

Wilson  Montero's profile picture
Wilson Montero4 days ago

I was using it in Ubuntu many months ago, it is better in Linux, it doesn't need vision.

れいちぇる_Web制作出身のAI・Webアプリ開発者's profile picture
れいちぇる_Web制作出身のAI・Webアプリ開発者4 days ago

プレビューから破壊的変更なしと明記して、ブラウザ・モバイル・デスクトップ向けのComputer UseをGAにしたのは、実運用で互換性を見通しやすくて大きいですね。モデルの能力だけでなく、画面操作の安定性が日々のエージェント運用を支えると思います。

M H Tawfik's profile picture
M H Tawfik3 days ago

When are we getting computer use inside Antigravity?

easyjackson's profile picture
easyjackson4 days ago

能替我点鼠标了,下一步能不能替我点掉那些订阅续费 🥲

Omni 🪽's profile picture
Omni 🪽4 days ago

Finally, someone who can close my 47 open tabs

Oyu|Grok Bot & AI's profile picture
Oyu|Grok Bot & AI4 days ago

no breaking changes from preview is honestly the best part. means anyone who built on it early doesn't have to redo anything

⚔️𝕯𝖎𝖌𝖎𝖙𝖆𝖑 👹 𝕽𝖔𝖓𝖎𝖓`鬼` (クラッシュ・オーバーライドX)'s profile picture
⚔️𝕯𝖎𝖌𝖎𝖙𝖆𝖑 👹 𝕽𝖔𝖓𝖎𝖓`鬼` (クラッシュ・オーバーライドX)4 days ago

awesome glad to compare it to my own and see i was only missing one small line of code really amazing work from @GoogleAIStudio

jimerrors's profile picture
jimerrors4 days ago

No breaking changes is the part I actually care about, my desktop agent can finally stay put.

凉杰's profile picture
凉杰4 days ago

Gemini电脑操作已GA,就怕弹窗死循环搞出误删。

Marco Kelly's profile picture
Marco Kelly3 days ago

this is awesome. i really want an agent that handles all the boring browser stuff for me. does anyone know if this works with antigravity yet

Moriarty's profile picture
Moriarty4 days ago

Hope it types cleaner than Codex computer use. Mine doubled every keypress, 2–3 letters per tap 🫠

ネム💤AIの料金と新機能's profile picture
ネム💤AIの料金と新機能4 days ago

GAになっても料金は別枠じゃなく、公式の料金ページだとComputer Useは無料枠では使えず、有料で通常トークン課金。推奨のGemini 3.8 Flashは入力$0.75/出力$3.75(100万トークンあたり、12/31まで)。こういう料金の線引き、公式ページで追ってます💤

Viber · fireply.ai's profile picture
Viber · fireply.ai4 days ago

flights from berlin to istanbul by script is a fun first test. i'd have a price alert bot on this by sunday

Elena AI's profile picture
Elena AI4 days ago

Computer use going generally available with no breaking changes is exactly what builders needed production ready stability for browser mobile and desktop agents is huge

zrkaria's profile picture
zrkaria3 days ago

Yes, but the gemini is too slow these two days. What can you do?

krsuncom's profile picture
krsuncom3 days ago

GA라고 바로 붙여보려는 사람은 이거 하나는 챙기자. 문서상 프롬프트 인젝션 감지는 옵트인이고 기본값이 false다. enable_prompt_injection_detection을 직접 켜야 스크린샷 속에 숨겨둔 "이전 명령 무시" 같은 지시를 잡아서 실행을 막아준다. Gemini 3.5 Flash-Lite 이상에서 지원. 그리고 좌표는 1000x1000 정규화 값으로 오니까 실제 뷰포트 크기로 환산해야 하고, 응답에 require_confirmation이 붙으면 사용자 확인 받는 처리도 클라이언트 쪽에서 직접 만들어야 한다.

Ansh's profile picture
Ansh4 days ago

@grok what does this mean, can i it use my computer like i do on gemini app ro google antigravity?

SiMelt's profile picture
SiMelt4 days ago

“全面可用、没有破坏性变更”这句对开发者很关键:意味着之前在预览版里跑通的浏览器、移动端、桌面代理流程,不用重写就能进生产。 补充一个落地的角度:电脑操作类代理一旦能点按钮、填表单,真正的门槛就从“能不能跑”变成“敢不敢放手”——付款、发消息、改设置这类操作,必须有可审计的操作记录和人工确认点,否则省下来的时间可能变成追回错误的时间。 你会先让这类代理替你做哪类重复操作:填表整理,还是跨应用搬运信息?

Tint's profile picture
Tint4 days ago

You guys see what’s going on with @PlaygroundPadHQ ?? Pretty amazing stuff.

Related Videos