GEMINI LABJP
NANOLITE — Nano Banana 2 Lite is here: Google's fastest and most cost-efficient Gemini Image model, made for running lightweight image generation cheaplyOMNIFLASH — Gemini Omni Flash is in public preview, a natively multimodal model that lets enterprises and developers build custom, dynamic video workflowsAGENTS — Managed Agents expand with background: true for async server-side runs and polling, remote MCP server integration, and refreshing credentials across interactionsMEMORY — The Memory Bank IngestEvents API is generally available, decoupling event ingestion from memory generation so you can stream content continuouslyTHROUGHPUT — Provisioned Throughput now lets you submit up to seven pending orders for the same model and regionDEPRECATE — Image generation models shut down on August 17, and the Grok 4.1 family on the Gemini Enterprise Agent Platform on August 20NANOLITE — Nano Banana 2 Lite is here: Google's fastest and most cost-efficient Gemini Image model, made for running lightweight image generation cheaplyOMNIFLASH — Gemini Omni Flash is in public preview, a natively multimodal model that lets enterprises and developers build custom, dynamic video workflowsAGENTS — Managed Agents expand with background: true for async server-side runs and polling, remote MCP server integration, and refreshing credentials across interactionsMEMORY — The Memory Bank IngestEvents API is generally available, decoupling event ingestion from memory generation so you can stream content continuouslyTHROUGHPUT — Provisioned Throughput now lets you submit up to seven pending orders for the same model and regionDEPRECATE — Image generation models shut down on August 17, and the Grok 4.1 family on the Gemini Enterprise Agent Platform on August 20
Articles/Gemini Basics
Gemini Basics/2026-05-07Beginner

Putting Gemini on Your Desktop and Phone as a PWA — How It Compares to a Browser Tab in Daily Use

How to install gemini.google.com as a Progressive Web App on macOS, Windows, iPhone, and Android — and what I actually measured after a week of comparing PWA usage to keeping it as a Chrome tab.

Gemini75PWA2DesktopiPhone2Android10Workflow4

One morning I needed Gemini to review some code, and finding gemini.google.com among the twenty-something tabs in Chrome took almost ten seconds. Once is nothing, but doing it dozens of times a day quietly drains your focus.

Splitting Gemini out into a Progressive Web App in its own window made that small friction disappear. One click on a Dock icon and the conversation is right there. The more I looked into it, the more I noticed real differences between using Gemini in a browser tab and using it as a PWA — including some I did not expect, like battery drain and how the iOS keyboard behaves.

This article walks through how to install gemini.google.com as a PWA on desktop, iPhone, and Android, and shares what I actually measured during a week of comparing the two setups. The piece is written for non-developers as well as developers.

Why bother making Gemini a PWA — four differences I noticed in a week

I assumed it would just be a nicer-looking icon in the Dock. It turns out the behavior diverges from a browser tab in several places.

The first difference is escaping tab management. Chrome's session restore is helpful, but when Gemini is mixed with thirty other tabs, a careless restart can bury a long conversation. A PWA runs as its own process, so it is not affected when you switch Chrome profiles or close all tabs.

The second is fewer keyboard shortcut collisions. Inside a PWA window, browser shortcuts like Cmd+L (focus address bar) and Cmd+T (new tab) are disabled, which means I no longer accidentally close the Gemini tab with Cmd+W while editing a long prompt.

The third is independent memory pressure. I will share the actual numbers below, but the practical benefit is being able to restart Gemini alone without restarting Chrome and losing every other tab's state.

The fourth is better keyboard behavior on iPhone. This was the biggest surprise. The same gemini.google.com page behaves better as a home-screen PWA than inside Safari — focusing the input does not push the cursor uncomfortably high on the screen when you are writing a long prompt.

Installing on desktop — Chrome, Edge, and Arc

Let's start with the most common case: Chrome.

Chrome (macOS / Windows / Linux)

  1. Open https://gemini.google.com in Chrome and sign in
  2. A small "install" icon appears at the right end of the address bar (a monitor with a downward arrow)
  3. Click it and choose "Install"
  4. The app is registered under /Applications/Chrome Apps/ on macOS, the Start menu on Windows, and ~/.local/share/applications/ on Linux

After installation you can pin it to the Dock or taskbar like any native app. On my Mac the small win is that Spotlight finds Gemini instantly when I type gemini, so launching from a hotkey launcher no longer requires hunting through tabs.

Microsoft Edge

Edge works the same way but ships with a slightly more polished install dialog.

  1. Click the "Install app" icon at the right edge of the address bar
  2. Choose "Install"
  3. After installation, you can pick whether to pin to the taskbar or pin to Start

PWAs installed from Edge appear in Windows Settings → Apps alongside other applications, which makes them easier to manage on a corporate device.

Arc, Brave, Vivaldi, and other Chromium browsers

Chromium-based browsers expose almost the same UI. Arc has its "Sites" feature that creates a PWA-like dedicated window, but it is not strictly the same as a true PWA install — Service Worker behavior occasionally diverges. I use Arc as my main browser, but for Gemini I install the PWA from Chrome.

Safari is not an option (on macOS)

macOS Safari does not currently support installing websites as PWAs. You can "Add to Dock", but it just opens as a regular Safari tab. If you want the PWA experience on a Mac, install through Chrome or Edge.

Installing on iPhone — and a quirk to watch for

On iPhone you install through Safari. Adding to the home screen from Chrome only creates a bookmark, so make sure to use Safari.

  1. Open https://gemini.google.com in Safari
  2. Tap the share button (square with an upward arrow)
  3. Scroll down and choose "Add to Home Screen"
  4. Confirm the app name and tap "Add"

When you launch from the new home-screen icon, Gemini opens in a dedicated view without Safari's address bar or tab strip. Two practical differences from Chrome on iOS: the input scroll behavior is more stable when focused, and it shows up as its own card in the app switcher.

There are caveats. Push notifications are technically supported from iOS 16.4 onward, but Gemini does not currently send any, so that benefit does not apply. Some Apple ID interactions and Shortcuts integrations also behave inconsistently inside Safari-based PWAs.

Installing on Android

Android handles PWAs more cleanly than iOS, and the result feels closer to a native app.

  1. Open https://gemini.google.com in Chrome
  2. Tap the three-dot menu at the right of the address bar
  3. Choose "Install app" or "Add to home screen"
  4. Confirm the name and tap "Install"

Android registers the PWA as an APK, which means it appears in the app drawer, the recent apps list, and Split-Screen multitasking. If your Google account is already signed in to the browser, you can resume conversations immediately after launching.

A note for Pixel users: the on-device Gemini Assistant is a separate product from the PWA. The assistant excels at voice and system actions, while the PWA is better for long conversations or Deep Research. Using both side by side is the most efficient setup.

Tab vs PWA — what I actually measured for a week

Here is the part I personally found most interesting. Same MacBook Pro (M2 Pro / 16GB / macOS 15.4), same workload, switched between tab and PWA usage across a week. Numbers below are not extrapolations, just what I saw in Activity Monitor and Battery preferences.

Memory consumption

With five long conversations open in parallel:

  • As Chrome tabs: Gemini's renderer process is folded into Chrome, with the combined Chrome footprint around 2.4 GB
  • As a PWA window: a dedicated process at around 980 MB (Chrome continues to run separately)

The PWA's footprint is more legible because it is isolated. More importantly, you can restart Gemini without restarting Chrome — useful during long working sessions.

Time from launch to last conversation

I timed how long it took from typing gemini in Spotlight to seeing the most recent conversation:

  • Hunting for the tab in Chrome: as much as 10 seconds when many tabs are open, around 3 seconds in light sessions
  • Launching the PWA: average 1.8 seconds, never longer than 2.5 seconds

Subjectively it feels three to five times faster. Twenty launches a day adds up to about three minutes saved daily, or around an hour a month.

Battery drain (M2 Pro, idle)

I left Gemini open for an hour and looked at battery usage:

  • Chrome tab (with 10 other tabs open): around 7% drained
  • PWA window only: around 4%

Most of that gap comes from Chrome's other tabs, not Gemini itself. Still, when I am only using Gemini for a stretch of time, the PWA is the more efficient choice.

Shortcut safety

Inside a PWA, Chrome shortcuts like Cmd+T and Cmd+Shift+T are disabled, so I cannot accidentally lose a conversation tab. The trade-off is that Cmd+L does not focus an address bar, so copying the conversation URL takes one extra click via the in-window menu.

Three small troubles I ran into

PWAs are convenient, but there are a few first-time pitfalls. These are the ones I actually hit.

The install icon does not appear in Chrome

If you do not see the install icon in the address bar, you are most likely either signed out or browsing in an incognito window. Sign in to gemini.google.com properly and scroll the page a bit; the icon usually shows up after a moment.

If it still does not appear, your Chrome may be out of date. Open chrome://version/ to check, and update if needed. I once spent twenty minutes debugging this on an old beta build before realizing the version was the issue.

iPhone PWA hangs on the splash screen

A common cause is stale Safari cache. Go to Settings → Safari → "Clear History and Website Data" and try again. Clearing the cache will sign you out of Google, so prepare your two-factor method before doing this.

Only the PWA gets signed out

Even though the PWA window and Chrome share cookies, the PWA occasionally loses its signed-in state on its own. Just signing back in inside the PWA fixes it. Because this can happen mid-conversation, I make a habit of sending long prompts in stages so the history is preserved.

Related reading

If you want to go further with PWAs and AI:

What I would like you to do next

If you have read this far, please install gemini.google.com as a PWA on the device you are reading this on. It takes about three minutes. Tomorrow's first conversation will start a little more smoothly than today's.

Thank you for reading.

Share

Thank You for Reading

Gemini Lab is ad-free, supported entirely by members like you. We publish practical guides daily with implementation code, benchmarks, and production-ready patterns. If you've found it useful, we'd love to have you on board.

  • Copy-paste ready implementation code
  • New advanced guides published daily
  • $5/mo or $10 for lifetime access
View Membership →

If you found this article helpful, a small tip ($1.50) would mean a lot to us. Your support helps keep this site ad-free and covers server and hosting costs.

Related Articles

Gemini Basics2026-03-20
How to Use the Gemini App: A Beginner's Guide to AI on Your Smartphone
Master the Gemini app on iPhone and Android. Learn text conversations, image recognition, voice input, and Google service integration for everyday AI use on your smartphone.
Gemini Basics2026-07-18
"No Watermark Detected" Doesn't Mean It Isn't AI — The Asymmetry of SynthID
Images generated with Gemini carry a SynthID watermark. But a positive result and a negative result don't carry the same weight, and that asymmetry changes how you should track provenance.
Gemini Basics2026-06-22
Putting Gemini image generation to work: from prompt design to thumbnails generated from video
A practical playbook for running Gemini image generation as a repeatable workflow instead of a lucky dip. From decomposing prompts into reproducible parts to the video-to-image automation unlocked by the Nano Banana 2 GA, with working code, a pre-publish quality gate, and a design that survives preview shutdowns.
📚RECOMMENDED BOOKS
Build a Large Language Model (From Scratch)
Sebastian Raschka
LLM Dev
Prompt Engineering for LLMs
Berryman & Ziegler
Prompting
AI Engineering
Chip Huyen
AI Eng
* Contains affiliate links
See all →