TL;DR
As of June 2026, Gemini runs the Gemini 3.1 Pro model and is the default assistant on most modern Android phones (it replaces Google Assistant, with the final mobile Assistant shutdown targeted for March 2026). The genuinely useful mobile tasks are camera translation, message triage by voice, quick drafting, visual ID, and one-line task capture. Gemini Live (continuous voice plus live camera or screen share) is now free for every Android user — no Google AI Pro subscription needed — on any phone with 2 GB of RAM running Android 10 or later. Long-form writing, screen sharing with another person, and anything sensitive still belong on a laptop.
What this tutorial solves
Gemini on Android is marketed as the answer to everything from cooking to coding. In practice, about five specific tasks make it worth keeping on your home gesture; the rest is faster done by hand. This guide names the tasks where Gemini on Android genuinely beats typing on a phone, gives you the exact gestures and settings, and is honest about the cases where you should just open your laptop.
Who this is for
Android users on Pixel, Samsung Galaxy, or any phone running Gemini as the default assistant. It is most useful if you want your phone to save time rather than become another notification source. iOS users get the standalone Gemini app and free Gemini Live, but the system-level assistant integration (power-button launch, default-assistant hooks) is shallower, so a few steps below will not apply.
What changed in 2026
- Gemini is the default assistant. Google is retiring Google Assistant on mobile, with the final shutdown targeted for March 2026. Phones with under 2 GB of RAM or Android 9 and earlier stay on the old Assistant. (9to5Google)
- Gemini Live is free for everyone. The live camera and screen-share modes used to require the old Gemini Advanced tier. As of 2026 they are free for all Android and iOS users on any device with 2 GB of RAM and Android 10+. (Android)
- The model is Gemini 3.1 Pro. Paid Google AI Pro ($19.99/month, formerly “Gemini Advanced”) adds the full 1M-token context window, Deep Research, and Gemini across Gmail/Docs/Sheets. The free app still runs Gemini 3.1 Pro for everyday queries.
What works well vs. what to skip
| Task | Best input | Why it wins on phone | Skip if |
|---|---|---|---|
| Translate a menu or sign | Camera | Live OCR beats retyping into Translate | Legal/medical text — verify with a second source |
| Triage the last hour of messages | Voice | One spoken prompt vs. scrolling | You need to act on each, not just skim |
| Draft a short reply | Voice | Faster than two-thumb typing | The reply is long or formal |
| Identify a plant, part, or object | Camera | Live camera beats reverse image search | You need a verified ID (mushrooms, dosage) |
| Capture a task or reminder | Voice | ”Add buy stamps tomorrow” beats unlocking + typing | Multi-field entry with dates and notes |
| Write 500+ words | — | Small screen, no real editing | Always — use desktop |
| Share a screen with a person | — | Gemini shares with the AI, not a human | Always — use a call/meet app |
Before you start
- Set Gemini as your default digital assistant. Go to Settings, then Apps, then Default apps, then Digital assistant, and pick Gemini. About half the gestures will not fire until you do this.
- Review Gemini Apps Activity (in the Gemini app, tap your profile, then Gemini Apps Activity). By default it retains your chats; Live audio, screen shares, and video are stored there but, per Google, are not used for product improvement. Turn retention off if you prefer.
- Test the launch gesture in a quiet room. Long-press the power button, or say “Hey Google,” and confirm Gemini responds before you rely on it in noise or while driving.
- For driving, pair Android Auto in advance. Configuring while driving is dangerous, and most options require unlocking the phone.
Step by step
- Make Gemini the default assistant. Settings, then Apps, then Default apps, then Digital assistant. Confirm by long-pressing the power button or saying “Hey Google” and seeing Gemini respond.
- Practice “What’s on my screen?” Long-press power, then ask about the visible content. Useful for translating an on-screen menu, summarizing an email, or pulling an address out of a screenshot. Remember Gemini reads the entire visible screen, so close sensitive apps first.
- Use the camera for the physical world. Open Gemini Live, tap the camera icon, and point at the object: “What is this?” or “Translate this sign.” Best for foreign-language signage, plant or part ID, and quick OCR of receipts or whiteboards.
- Dictate full thoughts, not commands. “I am walking, summarize my unread messages and tell me which need a reply” produces a usable result. “Summarize messages” produces something generic.
- Set up hands-free for driving while parked. Gemini reads notifications and lets you reply by voice through Android Auto. Send one test message and confirm it before trusting it with real ones.
- Decide your privacy posture. Review Gemini Apps Activity and turn off retention if you do not want chats kept. Note that Workspace integrations (Gmail, Docs, Calendar) carry separate data terms.
Gemini Live: the one feature worth learning
Gemini Live is continuous voice with optional live video. Long-press power to launch Gemini, tap the Live icon, then choose Camera or Share screen with Live. You can interrupt mid-sentence, pause or stop sharing, and switch between front camera, rear camera, and screen on the fly.
Two concrete uses that justify it:
- Walking commute: natural back-and-forth (“what’s the news, then summarize my unread Slack”) instead of one-shot commands.
- Live troubleshooting: point the rear camera at an appliance error code or a cable and ask what it is and what to do next.
It is free on any Android phone with 2 GB of RAM and Android 10 or later, so there is no paywall reason to skip it.
First-run exercise
- Pick the one task from the table above that fits your day best.
- Run it twice — once with voice, once typed on the small screen. Time each.
- Note which input mode wins and by how much. Voice typically wins by 2-3x on these.
- Repeat for the next task. After three or four, you will know your personal Gemini-on-Android shortlist.
Quality check
- Voice dictation: read or listen to the transcription before sending. Noisy environments produce silent errors.
- “What’s on my screen?”: Gemini is reading everything visible — close banking and messaging apps first.
- Translation: spot-check key phrases against a second source when the stakes are real (medical, legal, anything you sign).
- Task creation: open Google Tasks afterward and confirm the item landed in the right list.
Common mistakes
- Using Gemini for tasks that need precise text entry on a small screen. Type it, or wait for desktop.
- Not setting Gemini as the default assistant — half the gestures stay dead.
- Forgetting “What’s on screen?” reads the whole screen. Close sensitive apps first.
- Long voice prompts in noisy places. Transcription errors compound and ruin the output.
- Trusting Gemini’s translation of safety-critical text without a second source.
- Leaving Gemini Apps Activity on by default without checking what is being retained.
Advanced tips
- Write down your top three voice prompts and practice the phrasing. Spoken prompts have a different rhythm than typed ones.
- Use Gemini Live for walking commutes instead of one-shot commands.
- For task capture, just say “Add to my tasks: buy stamps.” Gemini routes it without opening the app.
- Break multi-step requests apart. “Find the closest pharmacy and add it to my route” works; “find a pharmacy, check hours, route there, then text mom I’ll be late” tends to drop steps.
FAQ
- Do I need to pay for Gemini Live’s camera and screen share? No. As of June 2026 it is free for all Android and iOS users on any device with 2 GB of RAM running Android 10 or later. Google AI Pro ($19.99/month) adds Deep Research, the full 1M-token context, and Workspace integration, but not the Live camera itself.
- Does Gemini replace Google Assistant? Yes, on most modern Android. Google is retiring mobile Assistant with the final shutdown targeted for March 2026. Phones under 2 GB of RAM or on Android 9 and earlier keep the old Assistant.
- Which Gemini model runs on my phone? Gemini 3.1 Pro as of June 2026. The free app uses it for everyday queries; Google AI Pro raises limits and unlocks the full 1M-token context window.
- Does it work without internet? No. Almost everything — Live, camera, translation, summaries — requires cloud calls. On-device assistant fallbacks are limited.
- What about Samsung’s Bixby? Samsung lets you set Gemini as the default assistant. Bixby coexists for a few device-specific tasks, but most users default to Gemini.