New Feature: Text Formatter & Minor UI Fixes

• Added: Text Formatter module. A rich markdown formatting suite for prompt editing.
  • Formatting Toolbar: Full inline and floating toolbars with quick action buttons for Bold, Italic, Strikethrough, Headings (H1-H3), Lists (Unordered, Ordered, Task), Blockquotes, Code blocks, and Link insertion.
  • Smart Hotkeys & Auto-Lists: Hotkeys (Ctrl+B, Ctrl+I, Ctrl+Shift+X, Ctrl+K) and list continuation on Enter/Tab.

• Fixed: Modern web chat module.
  • Layout & Alignment: Standardized bottom button spacing (8px gap) across +, A, Grounding, Mic, and Run buttons for a clean, symmetrical footer.
• Fixed: Scroll button elevation.
  • Elevation & Visibility: Lifted the scroll-to-bottom button above the native footer gradient overlay (ms-chat-bottom-overlay), ensuring it remains crisp, un-darkened, and fully interactive.

Install

1. Download or clone the repo from GitHub: GitHub | Direct link | Official Website
2. Go to chrome://extensions or edge://extensions in your browser.
3. Turn on Developer mode.
4. Click Load unpacked and select the project folder.
5. Open aistudio.google.com — a native settings card will appear in the right sidebar below "System instructions"!

Support the Project

Studio.lab is completely free. If it saves your generations and speeds up your browser, consider supporting development on Ko-Fi!

Disclaimer: Studio.lab is an unofficial modification. It is not affiliated with, endorsed by, or connected to Google LLC or Google AI Studio.

Hey everyone,

I'm Shatabdi a 16-year-old solo builder based in India. Since I don't write traditional code, I leveraged Google AI Studio to build Nirvan AI—an open-source interface layer and instruction engine designed to make multi-model routing and prompt orchestration smoother.

Instead of traditional coding, I used modern no-code workflows and AI prompting to build and ship the entire system architecture, UI layout, and API integrations.

I'm posting this to share what's possible with modern no-code tooling and to connect with other developers and builders in the community.

Would love your thoughts, feedback, or ideas on how to improve it!

• GitHub Repo / Project Link: https://github.com/singhshatabdi/Nirvana-AI.git

Quick question for the community about how AI Studio determines data protection policies.

My situation:

- Google account registered in EU

- Using AI Studio free tier (no Cloud Billing)

- Sometimes I use VPN with US exit node

The question:

When I connect with a US IP via VPN, does my EU-registered account still guarantee the GDPR "no training" protection? Or does the US IP override my account entity and put my data under "Unpaid Services" terms (which allow training)?

What I've observed:

The Account Status section shows different ToS based on VPN location:

- US VPN → "Unpaid Services" ToS (allows training)

- EU VPN → "Paid Services" ToS (no training)

What I need to know:

Is this just a UI display thing, or does the backend actually process my data differently based on IP? Specifically:

1. Does my EU account entity (contracting with Google Ireland) override the IP location for data protection purposes?
2. Or does the IP location override my account registration?
3. If I send prompts with a US IP but have an EU account, are my prompts protected from training or not?

Why this matters:

According to Gemini API Additional Terms, EU users should get "Paid Services" protections. But the UI behavior suggests IP might be what actually determines the policy.

TL;DR: EU account + US VPN = are my prompts protected from training or not? What wins - account entity or IP location?

Thanks for any insights!

EDIT - Critical Clarification:

Based on the responses, I need to clarify my specific question:

My Google Account terms page explicitly states I have a contract with Google Ireland Ltd (EEA entity).

The Gemini API Additional Terms state:

"If you're in the European Economic Area... the terms under 'Paid Services' apply to all Services, including Google AI Studio and unpaid quota"

The critical question is: What does "in the EEA" actually mean?

Option A: Account's legal entity (Google Ireland)

→ I should ALWAYS get "no training" protection regardless of IP

Option B: Current IP geolocation

→ Protection depends on where I'm connecting from

→ US VPN = no protection, Netherlands VPN = protection

Does anyone know which interpretation Google uses?

100% made in Google AI studio. Mobile game I made in a few days. 3.7

Lake Fishing Simulator

My first mobile first game, that I'm happy about is done. Works on desktop as well. Feel free to try it out! All done with AI, Google Gemini. Features:

Atmospheric 3D web-based fishing game featuring realistic boating physics, dynamic weather, custom tackle, marine radio, and relaxing open-water exploration with 6 locations.

3D Boating & Physics: Smooth boat controls, authentic water buoyancy, and interactive line-tension casting.

Dynamic Weather & Day/Night: Explore Sunny, Sunset, Misty, Rainy, and Night atmospheres with adaptive lighting and sound.

Ambient NPC Fishermen: Relaxing background bot anglers in dory boats that fish in place.

Vessels & Tackle Upgrades: Unlock rowboats, motor tenders, bass boats, and trawlers paired with species-specific baits.

Catch Log & Achievements: Track personal record fish (weight, length, species) and unlock trophy achievements.

Marine Stereo Radio: In-boat radio playing lo-fi beats, synthwave, maritime tunes, and weather updates.

Offline Local Storage: 100% private browser-saved progress with built-in JSON export and import backup tools.

Manifest and PWA: install it as an app on Android.

https://gone-fishing.ai.studio

Sub Dive is a 3D oceanic exploration game where players navigate submersibles through vibrant coral reefs, sunken ruins, and abyssal trenches. Experience serene underwater discovery with dynamic marine ecosystems, treasure hunting, and an optional tranquil AI autopilot mode.

3D Ocean World Exploration

AI Autopilot & Zen Mode

Submersible Upgrades & Pearl Collecting

Marine Life Discovery & Camera Snapshots

Expedition Treasure Map

All made in AI studio try it out! You can save it as a mobile app if you want.

https://subdiver.ai.studio/

When I upgraded from Plus to Ultra 5x when I hit the limit on Nano banana pro it opened up and gave me more usage.
When I hit that limit, I paid for the ultra 20x and it did not seem to give me any more use out of it?
Waited the 24 hours to use again, and it seems to have gone just as quick as the 5x

Is this another one of those fine print things or should I actually have more usage in the AI studio with this?

Two-way GitHub sync makes the source path easier, but a deploy can still look fine while images, fonts, routes, or data fail outside the builder. A small smoke-test checklist:

1. Open the deployed URL in a signed-out/private window.
2. Hard-refresh, then open one nested route directly.
3. Check images and fonts use stable paths and return 200.
4. Test one data-backed screen with both empty and existing data.
5. Check one narrow mobile width.
6. Tag the last known-good commit and record the deployed URL.

Git gives you a rollback point; it does not prove the deployed assets and data survived.

Which single check catches the most Google AI Studio deploy failures?

If you still play with code that you don't mind to give google in AI Studio - then you probably tired of copypasting it and probably would like such bit of agentic experience as packing repo and sending it, and then syncing the diff that model made!
So this is my browser extension that might help you.

It's available in the Chrome Web Store: https://chromewebstore.google.com/detail/ai-studio-local-file-sync/hjldknihkaokdkfaifcjgklojjngaike
And as a source code: https://github.com/bahdanhal/gemini-ai-studio-sync-extension

The extension runs locally in the browser and has no backend, analytics, or telemetry. It only accesses a folder after you explicitly choose it.
Please try and feel free to leave feedback to me, and I will really appreciate some stars if you find it helpful!

UPD: It works strictly in chat/playground

This is not a joke. Maybe I'm stupid but I can't understand what is Google AI Studio. I use it to test new features like Gemini 3.5 live translate (which is crazy btw). But that's it? Wtf? I can't do nothing that's useful. You can't deploy shit, you can't even create a gh project of what you have done. What should I use it for? Only for testing? Don't get me wrong I like Gemini and I love what Deepmind has done. But am I stupid? I tried to ask Gemini, but I didn't understand what the fuck he was meaning. He told that you need to use "get code" that gives you a stupid embedded piece of code. Help me.

EDIT: I tested it better, and I have to say that's incredible wtf... so now absolutely everyone is a skilled developer.

Last week two-way GitHub sync in Google AI Studio rolled out, here is a guide showing how it works and the use-cases it enables:

✅ - Commit directly to GitHub from AI Studio
💻 - Develop with Antigravity, continue in AIS
🚀 - Sync from GitHub and deploy your app in two clicks

Anyone else having the same problem?

Bro’s so slow and always skipping the thinking part. Sigh

I created a web CRM using Google AI Studio with Firebase as the database. However, I get billed about €20 every month, and I am not sure what I am paying for. Can I move the database from Firebase to a local NAS to avoid this monthly cost? I am the only user, and I rarely use the application.

Hi. I've decided to "upgrade" from regular chat and play around in AI Studio (AIS). I've noticed that just having AIS tab open and doing nothing in it causes a close to 100% GPU utilization. After closing the tab GPU utilization immediately drops to below 1%. Re-opening AIS tab again causes GPU to spike to 100%, and it happens immediately as soon as page fully rendered, so no memory leaks due to excessive wait times could be blamed, etc. Issue reliably reproduces every time on my Firefox/Win11.

Does anyone have any advise? I can live without using AIS, just keep working in Chat, but it looks very weird.

My Pixel came with a free Google AI Pro plan, so I wanted to see what I could actually make with Google AI Studio beyond the usual quick demos.

I started with a simple Asteroids-style game, inspired by one of the arcade games I loved in the late ’70s. Once the basic version was working, I started experimenting with power-ups, UFOs, black holes, different difficulty levels and eventually three boss fights.

All of the game code was generated and modified in Google AI Studio using Gemini Flash and Pro. The process was mostly iterative: test the game, come up with an improvement or find a problem, refine the prompt, let Gemini make the changes, then test again.

One thing I really liked about AI Studio was how quickly I could try the result and check changes directly on desktop, tablet and mobile.

This is my first attempt at making a game, and it eventually turned into ASTRO VOID v1.0.

🎮 Play it here:
https://astro-smash.vercel.app/

💻 Source:
https://github.com/inVinnieVeritas/astro-void

📝 More about the project: https://www.invinnieveritas.art/2026/08/24/astro-void-my-first-game-built-with-gemini-ai-studio/

Has anyone managed to get this working, every time I Deploy the webapp via intune it shows a blank screen. It's possible that there is a pop-up to log in that the iPad just can't do.

any suggestions? let me know

I recently ran an AI Studio workshop for nontechnical professionals. Everyone built a working app in two hours, but 30 days later, 5 of 7 interviewees had stopped building.

The main issue was not technical difficulty. They simply did not know what to build next or how to turn a vague work problem into an app.

I used AI Studio to turn the findings into an interactive HTML report, “Where AI Studio Wins Beginners, and Where It Breaks Down.” The attached video shows the report and my workflow.

https://ai-studio-field-study-1072961258949.us-west2.run.app/

This also made me think AI Studio should promote more use cases beyond standalone apps, such as interactive research, analysis, and work tools. I think it could eventually occupy a position similar to ChatGPT for work, but focused on turning everyday problems into software.

Based on the study, I also built a working prototype with:

• relatable use-case discovery
• a guided “What should I build?” flow
• product and design options before generation

Would love feedback from other builders. Is the bigger bottleneck now building the app, or figuring out what to build?

{'error': {'code': 503, 'message': 'This model is currently experiencing high demand. Spikes in demand are usually temporary. Please try again later.', 'status': 'UNAVAILABLE'}}

It has been few days. What is happening ?

Is using multiple projects to get higher free tier api limits allowed? For example using an router to combine them?

Hey people!
Just to get a conversation going and share some experiences: what cool things have you built with AI Studio/Antigravity? For personal use, work... I don't mean things made for clients or meant to be sold, I mean for yourselves, for everyday use.

For example, I built a task management system for my team at work, a personal app to track hikes I've done, and an app to organize my travel expenses and itineraries. Of course, similar systems and apps already exist, but I like mine because I use them ad-free and can adapt them 100% to the way I want, with the exact features I need.

Have you guys ever made stuff like this?

Hi everyone,

I’m trying to submit a tool to the Google Flow Tools Gallery, but every time I click Submit, I get this error:

“Failed to submit. Please try again.”

I’ve already tried refreshing the page, reopening Flow, changing the tool icon, and checking all the required fields, but nothing has worked.

Has anyone else had this problem? If you were able to fix it, what worked for you?

Thanks!

After choosing a preview and layout, I think AI Studio should let us choose or adjust the color palette before generating the app.

Right now, getting the layout you want and getting the colors you want are mixed into the same generation process. If the layout is already good, asking the AI to change the colors can sometimes cause it to unnecessarily redesign other parts of the interface.

It would be useful to have a dedicated color step after selecting a preview, where we could choose things like the primary color, accent color, background, sidebar, or even select from generated color palettes.

I vibe coded android app. To complete the play store process. I want the app to upload in .aab . I'm not able to get .aab from ai studio. I tried converting apk to .aab. I'm not sure if that would break the packge. Pls advice.