Google Gemini in 2026: Pricing, Free Plan, Models and Features

Google Gemini is Google’s AI assistant, free to use with a Google Account on the web, Android, iOS, Mac and Windows. Paid Google AI plans cost $4.99 to $199.99 a month in the U.S. and add video generation, higher limits and Gemini inside Gmail and Docs (Gemini subscriptions, checked Sept. 25, 2026). It makes the most sense if you already live in Google’s apps.

This review covers the state of Gemini as of September 25, 2026: models, features, the free plan, pricing, regional limits, privacy, and how it compares with ChatGPT and Claude. A lot has changed since the April 2025 version of this page: the Gemini 3 generation replaced the 2.x models, the single subscription became four tiers, video now runs on Gemini Omni, and NotebookLM is now called Gemini Notebook. Everything below comes from Google’s official pages. Génie Artificiel earns no commission on Gemini.

October 1, 2026 update: Google announced Gemini 4 Argon on September 30, with initial access restricted to selected cybersecurity partners. Wider access starting with paid API customers and Ultra subscribers has no firm date. See Gemini 4 Argon: access and announced pricing, based on the official announcement. The tables below retain their September 25 verification dates.

What Is Google Gemini?

• Maker: Google (models by Google DeepMind).
• Where to use it: gemini.google.com, Android and iOS apps, a Mac app (macOS 15 or later, since April 15, 2026) and a Windows app (Windows 10 or later, since Sept. 10, 2026) (Gemini release notes).
• Free: yes, with a Google Account.
• Paid plans (U.S.): Google AI Plus at $4.99/month, Pro at $19.99/month, Ultra at $99.99 or $199.99/month (Gemini subscriptions, checked Sept. 25, 2026).
• Age: Google AI plans require you to be 13 or older in most countries, and 18 or older in the European Economic Area, Switzerland and the U.K. (Gemini limits).

Which Gemini Models Are Available in 2026?

Google ships new models every few weeks. The app picks one by default, and a menu lets you switch models and thinking levels. Here is what the Gemini app offers as of this date (Gemini subscriptions and release notes, checked Sept. 25, 2026).

Gemini 3.8 Flash availability for Pro and Ultra subscribers comes from Google’s Sept. 2, 2026 announcement. Gemini Omni Flash was unveiled May 19, 2026 and rolled out to Plus, Pro and Ultra subscribers in the Gemini app and Google Flow (Google announcement).

The context window depends on your plan: 32,000 tokens without a subscription, 128,000 with Google AI Plus, and 1 million with Pro and Ultra, or about 1,500 pages of text or 30,000 lines of code according to Google (Gemini limits, checked Sept. 25, 2026). The often-quoted “million tokens” does not apply to the free plan.

What Can Gemini Do?

Deep Research: Sourced Reports

Deep Research drafts a research plan, browses hundreds of web pages (and, if you ask, your Gmail, Drive or uploaded files), then writes a report with its sources, usually in 5 to 10 minutes. It is available without a subscription, with lower limits, and Google warns it may be unavailable to free users at peak times. Pro and Ultra subscribers can run reports on the Pro model (Deep Research help, checked Sept. 25, 2026).

Canvas: Docs, Slides and Small Apps

Canvas opens a workspace next to the chat where you write or edit a document, a slide deck, code or a small web app that you can preview, debug and change in plain English. You can then turn a document into an Audio Overview or a quiz (Canvas help). Canvas is included in the free plan.

Gems: Custom Assistants

A Gem is a version of Gemini with standing instructions and reference files, such as an editor that follows your style guide or a meal planner. Gems you create on the web also show up in the mobile app and in the Gemini side panel in Google Workspace, and you can share them like a Drive file (Gems help). They are free, but they don’t work with Gemini Live.

Images, Video and Music

• Images: Nano Banana 2 creates and edits images from text or your photos, including infographics with legible text. The more precise Nano Banana Pro is for subscribers, with limited access in Flow for everyone else (Google AI plans).
• Video: with a Google AI plan, Gemini Omni creates a video with sound from text, photos or a video clip, then edits it as you chat. You can also record your face and voice to create a personal avatar and cast yourself in videos (English only). Uploading a video to edit isn’t available in the EEA, Switzerland, the U.K. or some U.S. states, and avatars aren’t available in the EEA, Switzerland or the U.K. (video help, avatar help, checked Sept. 25, 2026). For more control, the Google Flow video studio gives 50 free credits a day; it tops our top 10 AI video tools.
• Music: Lyria 3.5 generates full tracks, including instruments, vocals and lyrics.

Gemini Live, Gemini Notebook and More

• Gemini Live: voice conversation, now built into the chat, with screen or camera sharing on mobile. Free.
• Gemini Notebook: NotebookLM’s new name since July 16, 2026 (Google blog). You add your sources (documents, web pages) to question them or turn them into Audio Overviews. Paid plans raise the number and size of notebooks (Google AI plans).
• Import from another assistant: since March 26, 2026, Settings lets you import your preferences and chat history from other AI apps (release notes).

Agents: Gemini Spark and Gemini Agent

Gemini Spark is a personal agent that works in the background on your behalf: organizing files, building documents, running multi-step workflows in Google Workspace. It requires a Pro or Ultra plan, Keep Activity turned on and a personal account, and it is available wherever the Gemini app is, except the European Economic Area, Nigeria, Switzerland and the U.K. (Gemini Spark help, checked Sept. 25, 2026). Gemini Agent, which completes multi-step tasks on the web and asks for confirmation before critical actions such as a purchase, is for Ultra subscribers in the U.S., in English.

Gemini in Gmail, Docs and Google Workspace

This is what sets Gemini apart: it works inside the Google apps many people already use.

• Personal accounts: Google AI Plus adds Gemini in Gmail and Vids; Google AI Pro extends it to Docs and Sheets (Gemini subscriptions, checked Sept. 25, 2026). Even without a plan, you can connect Gmail, Drive or Calendar to the Gemini app and ask questions about them.
• U.S. extras: the AI Inbox in Gmail (Plus, Pro and Ultra), Daily Brief, a morning digest of your priorities drawn from Gmail and Calendar (all Google AI subscribers 18+), Personal Intelligence, which connects Gemini to Gmail, Photos, YouTube and Search (beta for Pro and Ultra since Jan. 20, 2026), and Chrome auto browse (Pro and Ultra, U.S. and India) (Google AI plans and release notes).
• Businesses: in Google Workspace plans, Gemini in Gmail is included from Business Starter, and Gemini in Docs, Sheets, Slides, Drive, Meet and Chat from Business Standard (Google Workspace pricing, checked Sept. 25, 2026). Google also sells Gemini Enterprise, an agent platform for teams, starting at $21 per seat per month for the Business edition (Gemini subscriptions).

Is Gemini Free?

Yes. With just a Google Account, the free plan includes Gemini 3.6 Flash, “varying access” to Gemini 3.1 Pro, Deep Research, Canvas, Gems, Gemini Live, image generation (Nano Banana 2), music generation and connected apps like Gmail or Drive (Gemini subscriptions and Gemini limits, checked Sept. 25, 2026). Three limits matter:

• the context window is 32,000 tokens, versus 1 million with Google AI Pro, so Gemini may miss parts of a long document;
• video in Gemini requires a paid plan, although the Google Flow studio gives 50 free credits a day (Flow help);
• usage limits aren’t published as a number of messages (see below), and Deep Research may be unavailable to free accounts at peak times.

Since May 17, 2026, Gemini’s limits are no longer message counts. They are compute-based: they depend on the model, the length of the chat and the features you use, refresh every 5 hours and are capped weekly. Google doesn’t publish a message count, only multipliers relative to the free plan (Gemini limits, checked Sept. 25, 2026). Image and video generation and Deep Research use up more of your allowance.

How Much Does Gemini Cost? Google AI Plans in the U.S.

Sources: Gemini subscriptions, Google AI plans, Gemini limits and, for free Flow credits, Flow help (checked Sept. 25, 2026). Paid plans include family sharing with up to five other people, at least for storage. In France, the same plans cost €4.99, €21.99, €99.99 and €219.99 a month.

Which plan should you pick? The free plan is enough to try Gemini, run research and generate images. Google AI Plus is the cheapest way to get video and Gemini in Gmail, plus 400 GB of storage. Google AI Pro is for long documents (1 million tokens), Gemini in Docs and Sheets, Gemini Spark or coding. Ultra, from $99.99 a month, is for heavy video and Deep Think use.

Students: eligible U.S. college students can get one year of Google AI Pro at no cost, and students in 140+ other countries get Google AI Plus. The offer must be redeemed by Dec. 31, 2026 and then renews at the standard monthly price unless you cancel (release notes, checked Sept. 25, 2026).

Where Is Gemini Limited?

Many Gemini features launch in the U.S. first, and some skip Europe entirely. As of Sept. 25, 2026:

• Launched for the U.S.: Gemini Agent (Ultra, English), the Gmail AI Inbox, Daily Brief and Personal Intelligence (beta); Chrome auto browse is in the U.S. and India. The release notes announce no wider rollout as of Sept. 25, 2026.
• Not in the EEA, Switzerland or the U.K.: Gemini Spark (also excluded in Nigeria), personal avatars, and uploading videos to edit with Gemini Omni (also unavailable in some U.S. states).
• Adults only: Gemini Omni video generation isn’t available to users under 18.

The core features (chat, Deep Research, Canvas, Gems, images, Gemini Live) are available wherever the Gemini app is.

Does Gemini Use Your Chats?

Worth knowing before you share personal or work data (Gemini Apps Privacy Hub, checked Sept. 25, 2026):

• While Keep Activity is on, your chats and files are saved, and Google uses them to improve its services, including to train its AI models. Some chats are reviewed by humans (including service providers); those excerpts are disconnected from your account and kept for up to 3 years, even if you delete your history.
• Activity auto-deletes after 18 months by default (you can pick 3 or 36 months, or never).
• If you turn Keep Activity off or use a temporary chat, conversations are kept for 72 hours and aren’t used to train models, unless you send feedback.
• Gemini inside Gmail, Docs or Drive: when you use it there with a Google AI plan, Google says it doesn’t use your content to train its models, unlike the Gemini app (Google Workspace help). Sharing Workspace data with the Gemini app through Personal Intelligence falls under the Gemini app rules.
• Work accounts: Google commits not to use Workspace data to train its models without permission (same page); Gemini Enterprise makes the same commitment (Gemini subscriptions).

Gemini for Developers: AI Studio and the API

Google AI Studio lets you try Gemini models and build prototypes at no cost, with limited quotas; the API is then pay-as-you-go. Gemini 3.8 Flash costs $0.75 per million input tokens and $3.75 per million output tokens through Dec. 31, 2026, then $1.50 and $7.50 from Jan. 1, 2027 (Google announcement, checked Sept. 25, 2026). Gemini 3.1 Pro is still in preview in the API (model list). Google AI Pro and Ultra subscribers also get higher limits in Jules, Google’s asynchronous coding agent, and Google Antigravity, its agent-first development platform.

Gemini vs. ChatGPT vs. Claude: Which Should You Choose?

Comparison based on official pages checked Sept. 25, 2026. All three companies change models and prices several times a year.

Sources: Gemini subscriptions, ChatGPT Plus and ChatGPT Pro tiers (OpenAI Help Center), ChatGPT features by plan, Claude pricing (checked Sept. 25, 2026). OpenAI paused new sign-ups and upgrades to the $200 Pro tier on Sept. 10, 2026; the $100 tier is still open. Sora’s shutdown is covered in our top 10 AI video tools.

On quality, the public Arena leaderboard, where people vote blind between two answers, had several Anthropic Claude models at the top of its text ranking as of Sept. 13, 2026. Google’s best model, Gemini 3.8 Flash, ranked 9th with a preliminary score, Gemini 3.1 Pro 15th, and OpenAI’s best (GPT-5.6 Sol) 18th. The gaps are small (13 points between 1st and 9th), and the ranking measures preference on short answers, not reliability on long documents.

In practice, Gemini stands out less for its model than for its ecosystem: Gmail, Docs, Drive, Android, included storage and video in the same subscription. ChatGPT offers a low-cost Go plan that may show ads, plus Codex and custom GPTs. Claude includes Claude Code and its Design, Slides and Docs tools from the Pro plan. For a European alternative, see Mistral Vibe (formerly Le Chat).

Who Is Gemini For?

Gemini is a good fit for:

• Gmail, Docs, Drive and Android users who want AI without switching tools;
• tight budgets: the free plan covers Deep Research, Canvas, Gems and images, and Google AI Plus adds video for $4.99 a month;
• video creators, with Gemini Omni and Google Flow in the same subscription;
• anyone analyzing long documents or codebases with Google AI Pro’s million-token window.

Look elsewhere if:

• you need to know exactly how many messages you can send: Google no longer publishes that;
• you live in Europe and want an autonomous agent like Gemini Spark: it isn’t available there;
• your data must never be used for training: turn off Keep Activity, use temporary chats or a work account.

For more everyday AI tools, see our top AI productivity tools and our piece on multimodal AI.

Google Gemini FAQ

What’s the Difference Between Google AI Plus, Pro and Ultra?

Plus ($4.99/month) doubles your limits and adds Gemini Omni video, Gemini in Gmail, a 128,000-token context and 400 GB. Pro ($19.99/month) quadruples your limits and adds Gemini 3.8 Flash, the million-token window, Gemini in Docs and Sheets, Gemini Spark, 1,000 Flow credits and 5 TB. Ultra ($99.99 or $199.99/month) multiplies Pro’s limits by 5 or 20 and adds Deep Think, plus Gemini Agent in the U.S. (U.S. prices checked Sept. 25, 2026).

Is Gemini Better Than ChatGPT?

Neither wins everywhere. On Arena’s text leaderboard as of Sept. 13, 2026, Google’s best model edges out OpenAI’s best by a few points, but both trail several Claude models. Gemini stands out for its Gmail and Docs integration, included storage and video from $4.99 a month; ChatGPT has its Go plan and its own tools such as Codex and custom GPTs.

What Is the Latest Gemini Model?

The latest announced model is Gemini 4 Argon, introduced September 30, 2026. As of October 1, access is restricted to selected partners; the announcement does not confirm general availability in the app. Gemini 3.8 Flash is part of the Pro and Ultra offering documented on September 25. Check Argon’s launch conditions before changing subscriptions.

Can Gemini Make Videos?

Yes, with any Google AI plan, starting with Plus at $4.99 a month. Gemini Omni generates videos with sound from text, photos or a video clip and edits them as you chat, and in the U.S. you can star in them with a personal avatar. You must be 18 or older.

Is There a Gemini Desktop App?

Yes. A Mac app (macOS 15 or later) launched April 15, 2026 and a Windows app (Windows 10 or later) on Sept. 10, 2026. Both open with a keyboard shortcut (Option + Space on Mac, Alt + Space on Windows), alongside the web version and the mobile apps.