The Complete Gemini AI Beginner's Guide 2026

Gemini is Google's AI, and in 2026 it is a lot more than the chat box at gemini.google.com. It is the assistant on Android phones, a native app on Mac and Windows, a side panel in Chrome, a helper inside Gmail and Docs, an agent called Spark that keeps working after you close your laptop, and the engine behind Google's image, video and music generators. If you already live in Google's world, Gemini is the AI that can actually see your email, your calendar, your photos and your files, and that is the thing no other assistant can match.

This guide walks the whole product the way it looks today. Which model to pick, which plan is worth paying for, where every feature lives, and the settings that decide what Google keeps about you.

Getting Started

You need a personal Google account. Sign in at gemini.google.com and you are chatting. Once you are set up, my Gemini secret codes list gives you 100 shortcuts to paste straight in, and Google's free AI tools covers the rest of the Google lineup. Everything below works on the free tier unless it says otherwise. Where you use it matters a little more than it used to, because the apps are not all identical:

Rick's tip

Set up the desktop app on day one. Being able to hit Option + Space (or Alt + Space) and ask a question without leaving what you are doing is the habit that makes Gemini stick. Same idea as Spotlight, except it can read your screen.

The Interface

The app was redesigned in 2026 and it is cleaner than it used to be. The left sidebar has New chat, Search chats, Library (your generated images, videos and music), Gems, Notebooks, and your recent conversations. Your profile picture and the settings gear sit at the bottom.

The prompt box is a pill at the bottom of the screen with a single + button. Everything lives behind it: uploads (photos, camera, files, Drive, Notebooks) and the Tools menu, which holds Guided learning, Deep Research, Canvas, Create music, Create video, Create image, and Personal Intelligence. Temporary chat is in the top right. The model picker is a dropdown at the top left of the prompt area, and the thinking level is a second menu under the model name.

You can upload up to 10 files per prompt: documents up to 100 MB each, videos up to 2 GB. Workspace files, PDFs, Word, Excel, CSV, text and Markdown all work. Free accounts get five minutes of total video and ten minutes of audio per prompt; Pro and Ultra stretch that to an hour of video and three hours of audio.

The Models, Explained

The picker shows three tiers with version numbers, and it is simpler than it looks.

Under the model name there is a Thinking level menu: Standard, Extended, and on Ultra plans, Deep Think. Extended thinking is free for everyone and makes the model reason longer before it answers. Deep Think is the heavyweight mode for problems that take minutes to solve properly, and it needs the Pro model plus an Ultra plan.

The rule that covers 90% of cases: Flash for everything, switch to Pro when the answer really matters, turn on Extended thinking for anything with multiple steps.

Rick's tip

Usage limits are compute-based now, not a daily message count. Every prompt costs an amount that depends on its complexity, the tools you used, and how long the chat is. Your allowance refreshes every five hours, with a weekly cap on top. A short question in a fresh chat is nearly free. A long chat with a video attached burns through it fast, so start a new chat when you change topics.

Pricing: What You Actually Get

Google sells four paid tiers, and the naming is confusing enough that most people pick wrong.

Which one:

Rick's tip

If you are a U.S. college student, Google is giving away 12 months of Google AI Pro free through December 31, 2026, verified through SheerID at gemini.google.com/students. It needs a payment method on file and renews at $19.99 unless you cancel, so put a reminder in your calendar for month eleven.

Deep Research

Deep Research is Gemini's research agent. You give it a question, it plans a search, reads dozens of sources, and comes back with a cited report, usually in five to ten minutes. It works on the free tier.

Where it lives: click the + in the prompt box, choose Deep Research, type your question, and hit Start research. You can attach your own files as sources, and you can click Sources to let it include your Gmail or Drive. When it finishes, click Open to read the report. From there you can turn it into a quiz, an infographic or an audio overview.

Rick's tip

Deep Research is the feature that rewards a specific question. "Research electric bikes" gets you a Wikipedia article. "Research which electric bike under $2,000 holds its value best after two years, using owner forums and resale listings" gets you something you could not have found yourself.

Gems

A Gem is your own version of Gemini with standing instructions. Make one for the tasks you repeat: a writing assistant that knows your voice, a study coach for one class, a meal planner that knows your allergies. You can attach files as its knowledge so it always has your reference material.

Where it lives: Gems in the left sidebar, then New Gem. Give it a name, write the instructions, add files under Knowledge if you want, test it in the preview, and Save. Gems are not available inside Temporary chats or Gemini Live.

Canvas

Canvas is a workspace next to the chat for documents, code and pages. Instead of regenerating a whole answer, you edit it in place. Highlight a paragraph and ask for a rewrite, change the tone of one section, or build a working web page and preview it.

Where it lives: on desktop, click Canvas under the prompt box. On mobile, tap + then Canvas. Everything saves automatically, and the Create button in the top right turns what you have into a quiz, an infographic, a web page or an audio overview. Share and export lives in the same corner.

Gemini Live

Live is the voice mode. Open the Gemini app on your phone and tap Live at the bottom, or swipe left, or say "Hey Google, let's talk Live." It is a real conversation: interrupt it, change the subject, think out loud.

The part that makes it different from every other voice assistant is that it can see. Tap the camera icon and point your phone at something, and it can tell you what it is, read the label, translate the sign, or walk you through fixing it. Tap Turn on screen sharing and it can look at whatever is on your screen while you talk. Live pauses the camera and screen the moment you lock the phone or leave the app.

Two things to know: Live does not work with Gems, and Google's own help pages still describe it as a phone feature, so treat it as mobile-first even though the web app has been getting a version of it.

Rick's tip

Live plus the camera is the best way to learn anything physical. Point it at the back of your router, the error on your dashboard, the plant that is dying, or the recipe you are halfway through, and ask.

Images: Nano Banana

Nano Banana is the name of Gemini's image models. There are three: Nano Banana 2 Lite, Nano Banana 2 (the default), and Nano Banana Pro, which paid plans can use to redo an image at higher quality.

Where it lives: open the sidebar and click Images, pick a template or type a prompt, or use Create image from the Tools menu. To edit something you made, open Library, click the image, click Chat underneath it, and describe the change. On a paid plan you can click More in the bottom right of any image and choose Redo with Pro. Downloads are 1K on the free tier and 2K on any paid plan.

Every image carries an invisible SynthID watermark, always. The visible spark icon in the corner can be turned off on the web under the settings gear, then Media watermark, but the invisible one stays.

Video: Gemini Omni

Omni is Gemini's video model. It makes 10-second clips from text, an image, or another video, and you can extend a scene in 10-second steps up to 40 seconds, set the first and last frame, draft at low resolution, and upscale to 4K. Video needs a paid plan (Plus, Pro or Ultra) and you have to be 18 or over.

Where it lives: open the sidebar and click Create video. Choose the aspect ratio, upload up to one video and five reference images if you want, and describe the shot. It takes a few minutes and you cannot use that chat while it works. Edits are conversational: ask it to remove an object, change the camera angle, or swap the scene, and it returns a new version.

For bigger video projects, Flow at labs.google/flow is Google's full video studio, and it comes with the paid plans with its own monthly credits.

Music

Gemini can write songs now. Choose Create music from the Tools menu, describe the genre and mood, pick vocal or instrumental and short or long, and it produces a finished track. Free for everyone, web and mobile.

Notebooks and Gemini Notebook

There are two notebook things and they sync with each other, which confuses everyone at first.

Notebooks inside the Gemini app live in the sidebar. Click New notebook, drop in chats, files and instructions, and every conversation you start inside it has all of that as context. This is how you keep a project or a class together.

Gemini Notebook is the standalone product at notebooklm.google (it used to be called NotebookLM). Upload your sources, and it answers only from them, makes study guides, and generates the famous podcast-style Audio Overviews. It can run code in a secure cloud computer now too. Anything you build in one shows up in the other.

Rick's tip

Audio Overviews also work straight from the Gemini app. Upload a document, ask for an audio overview, and you get a two-host conversation about it in three to five minutes. Great for the commute.

Personal Intelligence and Memory

This is the feature that separates Gemini from everything else. Personal Intelligence lets Gemini use your Gmail, Calendar, Photos, Search history, YouTube, Contacts and Workspace files to answer you. "What was the name of the hotel we stayed at in Lisbon?" "Find the invoice from the plumber." "Pull the flight details from my confirmation email into my calendar."

Where it lives: Settings and help, then Personal Intelligence, then Connected Apps, and pick which apps it may use. There is a Memory toggle in the same place that lets it remember your past chats. Both need you to be 18 or over on a personal account, and neither is available in the EU, UK or Switzerland. You can also switch Personal Intelligence off for a single chat from the Tools menu.

Google says it does not train on your Gmail or Photos through this feature, only on limited prompts and responses. Check the setting rather than assuming a default, because it can arrive switched on.

Daily Brief

If you turn on Personal Intelligence, Memory and your Workspace connection, Gemini builds you a morning briefing: Top of mind, FYI, and Looking ahead, assembled from your email, calendar and recent chats. It arrives as a notification and lives in the side panel, and you can ask for it in Live with "What's my daily brief?" U.S. only right now.

Connected Apps

Connected Apps are the services Gemini can act inside of. Google's own apps (Maps, YouTube, Photos, Home, Contacts, Messages) plus more than 70 third-party ones: WhatsApp, Spotify, Canva, Dropbox, GitHub, Instacart, OpenTable, Zillow, Ticketmaster, Zocdoc and more. In a chat, just name the app or type @ followed by the app name. Manage the list under Settings and help, then Personal Intelligence, then Connected Apps. A lot of the third-party ones are U.S. only.

Scheduled Actions

Tell Gemini to do something on a schedule and it will. "Every Monday at 8am, give me a summary of the week's tech news." "On the first of the month, remind me to review my subscriptions." It confirms the schedule, and you manage everything under Settings and help, then Scheduled actions, up to 10 active at a time. It needs Keep Activity turned on, and Google's plan comparison lists it as a paid feature, so expect it on Plus and above.

Gemini Spark: The Agent

Spark is the feature to understand if you pay for Pro. It is a personal agent that runs on Google's servers, so it keeps working after you close the laptop. Give it a task, a list of tasks, or a standing schedule, and it browses, books, fills forms, cleans your inbox, organizes Drive, and reports back. It can run up to 15 tasks at once and you can teach it reusable skills.

Where it lives: on the web, open the sidebar and click Switch to Spark, then Tasks. On the Mac app it is Spark in the sidebar. Pro or Ultra, 18 and over, personal accounts, and not available in the EU, UK or Switzerland. Google's own rule is worth repeating: never give Spark sign-in details or payment information. It pauses and asks you before anything sensitive.

Rick's tip

Start Spark on something boring and reversible, like sorting Drive or drafting replies, before you let it near anything with a checkout button. The point of the first week is to learn how it thinks.

Gemini in Chrome

The Gemini button at the top right of Chrome opens a side panel that can read the page you are on. Ask it to summarize an article, compare the products in your open tabs, pull dates into your calendar, or explain a contract clause. It is free on desktop in the U.S., UK, Canada, India and a growing list of countries, and on Android and iPhone in the U.S.

On Pro and Ultra in the U.S. there is Auto Browse: Gemini clicks through sites for you, fills forms, and completes bookings and orders, pausing for your confirmation before anything sensitive.

Gemini in Gmail and Workspace

Gmail gets Gemini for free: AI Overviews that summarize a long thread, Help me write, and suggested replies. Asking questions across your whole inbox, Proofread, and the AI Inbox that sorts and triages for you need a paid plan (Plus and up). The Gemini side panel inside Docs, Sheets, Slides, Drive and Meet needs Pro or Ultra on a personal account.

Study Tools

Gemini has a real study kit built in. Ask for a quiz, flashcards or a practice test on any topic and it builds one. Guided learning in the Tools menu turns it into a tutor that teaches step by step instead of handing you the answer. Storybook, under Gems, writes and illustrates a ten-page picture book from a single prompt. The student hub at gemini.google.com/students collects all of it.

Privacy: The Settings That Matter

Gemini saves your conversations to your Google account by default, and a sample of them can be read by human reviewers and used to train Google's models. Three settings decide how much of that happens.

1. Keep Activity. This is the master switch, and the history behind it is called Gemini Apps Activity. On the web: Settings and help, then Activity, then click On at the top and choose Turn off, or Turn off and delete activity. On the phone: Menu, your profile picture, Gemini Apps Activity, then On, then Turn off. The direct link is myactivity.google.com/product/gemini. With it off, new chats are not used for training or human review, and Google keeps them for at most 72 hours to run the service. The catch: Spark, Scheduled actions, Memory, Personal Intelligence and most Connected Apps all need it on.

2. Auto-delete. On the same Activity page, the default is 18 months. Set it to 3 months if you would rather your history did not pile up.

3. Temporary chat. Next to New chat on the web, or in the menu on mobile. Nothing from a temporary chat goes into your activity, memory or training, and Google deletes it after 72 hours. It cannot use Gems, Connected Apps or personalization, and on mobile it cannot generate images or take file uploads.

Rick's tip

One honest detail from Google's own privacy page: chats that were already pulled for human review are kept for up to three years even if you delete your activity, though they are disconnected from your account first. Turn Keep Activity off before the sensitive conversation, not after.

Prompting Gemini Well

Gemini responds best to prompts that read like a brief to a capable colleague. Three habits that change the output more than anything else:

25 Things To Try This Week

Where Gemini Is Strongest, And Where It Is Not

Strongest: anything that touches your Google life. Email, calendar, photos, Drive, Maps, YouTube. Deep Research is excellent. The image and video generators are among the best available. The free tier is generous. Spark is a real agent that works while you sleep.

Weaker: long-form writing with real voice still needs a heavy edit. The naming is a mess (two Ultra tiers, two kinds of notebooks, models that change version number every month). Usage limits are opaque, and a long chat with attachments can eat a five-hour window fast. Several of the best features are U.S. only or need a personal account, so work accounts and European users get a smaller product.

Your First Week

Pick one thing from the list above, open Gemini, and do it right now. That first result will tell you why Google put this in everything.