Google

Gemini

Provider

Google

Price

$0-99.99/mo

Platforms

Web, iOS, Android

Launched

Feb 2024

Updated July 2026

Gemini is Google’s AI assistant, built by Google DeepMind and rebranded from Bard in February 2024. After a rocky start — a $100 billion stock crash and a 2024 image-generation scandal — Gemini has staged a major comeback: the app passed 900 million monthly active users by Google I/O in May 2026, up from 400 million a year earlier, making it the second-largest AI assistant behind ChatGPT. Its biggest advantage is being woven into products billions already use — Search, Android, Chrome and Workspace.

What sets Gemini apart is very large context windows (1M tokens on the Pro models, with Gemini 3.5 Pro targeting 2M), strong native multimodality, and pricing that bundles Google One cloud storage with the paid tiers. Whether you’re processing thousand-page documents, generating video with Veo, or using AI across Gmail, Docs and Sheets, this guide covers the consumer app.

Quick stats

Pricing breakdown

Gemini’s pricing structure bundles AI capabilities with Google One storage, differentiating it from competitors offering standalone AI access. All prices are in USD.

Free — $0

The free tier is genuinely capable, not just a demo. You get:

• Gemini 3.5 Flash — Default model, the latest fast model (May 2026)
• Limited Gemini 3.1 Pro — Access to the flagship with usage caps
• Gemini Live — Real-time voice conversations in 45+ languages
• Basic image generation — Nano Banana / Imagen, with daily limits
• Deep Research — Autonomous multi-source research (limited monthly)
• File uploads and Google Workspace extensions — Gmail, Drive, Docs, Calendar, Maps, YouTube

Limitations: Spreadsheet and code file uploads restricted to paid tiers, shorter context window, no video generation, no custom Gems creation.

Worth it for: Casual users, students, anyone wanting a capable free AI assistant with Google integration.

Google AI Pro — $19.99/month

The sweet spot for regular users. Everything in Free, plus:

• Gemini 3.1 Pro with expanded limits
• 1M token context window — Process documents up to ~1,500 pages
• Higher Deep Research limits
• Custom Gems creation — Build unlimited personalised AI assistants
• Veo video generation and Nano Banana / Imagen image generation
• 2TB Google One storage — Shareable with up to 5 family members
• Priority access — Faster responses during peak times

Free for students: University students in the US, UK, Japan, Brazil, and Indonesia get Pro free for one academic year.

Worth it for: Regular users who hit free tier limits, anyone wanting video generation, professionals needing long-context processing, Google ecosystem users who’d benefit from 2TB storage.

Subscribe to Google AI Pro →

Google AI Ultra — $99.99/month (cut from $249.99)

For power users and professionals who need maximum capability. Everything in Pro, plus:

• Gemini 3 Deep Think — Advanced multi-hypothesis reasoning; early access to Gemini 3.5 Pro
• Highest prompt and Deep Research limits
• Full Veo video generation — Extended clips, higher quality
• Project Mariner — Agentic browser automation (US only)
• YouTube Premium — Ad-free YouTube, YouTube Music included
• 30TB Google One storage — 15× Pro allocation
• Priority compute — Fastest available response times

Worth it for: Researchers, consultants, heavy users who need Deep Think reasoning, anyone who’d pay for YouTube Premium separately anyway.

Learn more about Google AI Ultra →

Google Workspace — $14–22/user/month

As of January 2025, Google eliminated the separate Gemini add-on fee for business users. AI is now included in standard Workspace plans:

• Business Starter ($7.20/user/month) — Basic Gemini features
• Business Standard ($14.40/user/month) — Full Gemini in Gmail, Docs, Sheets, Slides, Meet
• Business Plus ($21.60/user/month) — Everything in Standard plus enhanced security
• Enterprise (custom pricing) — Advanced compliance, unlimited storage, custom data controls

Previously, Gemini for Workspace cost $20–30/user/month as an add-on. Now it’s bundled free, making Google’s enterprise offering significantly more competitive with Microsoft Copilot.

Models powering Gemini

Google ships Gemini on a Pro/Flash cadence plus a Deep Think reasoning mode. In the app you pick a model from a dropdown; here’s what’s current.

Gemini 3.1 Pro (February 2026)

The current generally-available flagship, with a 1M-token context window and strong multimodal reasoning. It’s the default Pro-class model on the paid tiers and the limited-access Pro model on Free. Gemini’s signature strengths — huge context and native multimodality — are at their best here.

Gemini 3.5 Pro (limited preview)

Announced at Google I/O on 19 May 2026, Gemini 3.5 Pro targets a ~2M-token context window and Deep Think reasoning. It slipped past its I/O date and, as of mid-June 2026, is still rolling out in limited preview rather than wide release, with early access for Ultra subscribers.

Gemini 3.5 Flash (May 2026)

The latest fast, low-cost model, shipped at I/O and the default on the Free tier. It is strong for everyday tasks and reportedly outscores the previous-generation Pro on some coding benchmarks.

Gemini 3 Deep Think

An extended-reasoning mode (Ultra) that explores multiple hypotheses in parallel for the hardest maths, science and reasoning problems, at higher latency and cost.

Gemini Nano

The on-device model runs locally on Pixel 9+ and Samsung Galaxy devices:

• No internet connection required
• Privacy-preserving (data never leaves device)
• Powers features like Call Notes, Smart Reply, and on-device summarisation

Features deep dive

Gemini Live voice mode

Gemini Live provides real-time conversational AI that feels genuinely natural. Unlike competitors that charge for voice mode, Gemini Live is free for all users (September 2024 expansion).

Key capabilities:

• 45+ languages supported with 10 voice options
• Natural interruption — Cut in mid-sentence, just like human conversation
• Adjustable speech rate — Speed up or slow down during conversation
• Emotionally responsive — Adapts tone based on conversation context
• Background operation — Continue conversations while using other apps

Limitations: Extensions (Gmail, Drive, etc.) aren’t accessible during voice conversations yet.

Deep Research

Google pioneered autonomous research in December 2024, inspiring ChatGPT’s competing feature. Deep Research autonomously browses hundreds of websites and your personal Workspace content to produce comprehensive reports.

How it works:

1. Submit a research question
2. Gemini creates a research plan (you can modify it)
3. Agent browses web and connected Google services
4. Produces a multi-page report with citations

Workspace integration: As of October 2025, Deep Research can search your Gmail, Drive, Docs, and Chat—combining public research with your private documents.

Monthly limits: 5 reports (Free), 20/day (Pro), 200/day (Ultra).

Image generation with Imagen and Nano Banana

Google’s Imagen models power Gemini’s image generation, producing high-resolution images with strong text rendering and character consistency. The Nano Banana image model (built on Gemini 3) went viral in September 2025, briefly pushing Gemini to #1 on both app stores, and its Pro version remains Gemini’s headline image tool.

Key features:

• Photorealistic or stylised outputs
• Excellent text rendering in images
• Character consistency across generations
• SynthID watermarking — Visible for free users, invisible for Ultra

The controversy: In February 2024, Gemini’s image generator produced historically inaccurate images (diverse Nazi soldiers, female popes) due to overzealous diversity prompting. Google paused people generation for six months, returning with improved guardrails in August 2024.

Video generation with Veo and Gemini Omni

Google’s Veo models create video clips with native audio — dialogue, sound effects and ambient noise generated together.

Capabilities:

• Short clips at 720p–1080p, with scene extension to longer durations
• “Ingredients to Video” — Upload reference images for style/character consistency
• Photo-to-video animation

Availability: a faster Veo on the Pro tier and the full model on Ultra. Video generation is not available in the EEA, Switzerland or UK due to regulatory restrictions.

Gemini Omni

Announced at Google I/O on 19 May 2026, Gemini Omni is Google’s unified multimodal video model — a single system that takes text, image, audio and video in one prompt and returns video, edited photos or a digital avatar, with shared context across every modality. Where Veo is a text-to-video generator, Omni collapses generation and editing into one model.

• Conversational editing — Generate a clip, then keep iterating in chat: change a shot, swap a prop or redo the camera move without starting over
• Character consistency — Faces, outfits and props stay consistent across shots, a known weak spot for rival video models
• Physics understanding — Improved intuitive grasp of gravity, kinetic energy and fluid dynamics for more realistic scenes

Availability: Gemini Omni Flash is rolling out to all Google AI Plus, Pro and Ultra subscribers globally through the Gemini app and Google Flow, and free on YouTube Shorts and the YouTube Create app. Flash caps clips at 10 seconds for now (a deployment choice, Google says, not a model limit). A higher-end Gemini Omni Pro has been teased without a release date. As with Veo, availability is subject to the same EEA/Switzerland/UK regulatory restrictions.

Canvas for collaborative editing

Canvas launched in March 2025 as Gemini’s answer to ChatGPT’s Canvas and Claude’s Artifacts—an interactive workspace for documents and code.

Writing features:

• Highlight text for targeted tone adjustments
• Export to Google Docs
• Convert to audio overviews (podcast-style with two AI hosts)

Coding features:

• Live code preview for React/HTML/CSS
• In-line editing and debugging
• Syntax highlighting for major languages

Gems for custom AI assistants

Gems are customisable AI personas with specific instructions, personalities, and knowledge. Think of them as Google’s answer to Custom GPTs.

Premade Gems:

• Brainstormer — Creative ideation
• Career Guide — Professional advice
• Coding Partner — Programming assistance
• Writing Editor — Content refinement
• Chess Champ — Chess analysis and training
• Learning Coach — Educational support

Custom Gems (Pro/Ultra): Create unlimited Gems with custom instructions and uploaded knowledge files. Shareable via Google Drive links since September 2025.

Extensions connect Google services

Gemini’s Extensions (now called Apps) integrate directly with Google services, enabling queries that span multiple data sources:

Available integrations:

• Gmail — Search emails, compose replies, summarise threads
• Google Drive — Find and summarise documents
• Google Docs — Edit and create documents
• Google Calendar — Check schedule, create events
• Google Tasks — Manage to-do lists
• Google Keep — Access notes
• Google Maps — Get directions, find places
• YouTube — Search and summarise videos
• Google Flights — Search flight options
• Google Hotels — Find accommodation

Example prompt: “Find the contractor bid from last month’s emails and add a meeting to discuss it on my calendar.”

File handling and code execution

Gemini supports extensive file uploads with generous limits:

Supported formats: PDF, DOC, DOCX, XLS, XLSX, CSV, Python, Java, and most image/video formats.

Size limits:

• Documents: 100MB per file
• Video: Up to 2GB
• Files per conversation: 10 (Free), more on paid tiers

Code execution: Gemini runs Python in a sandboxed environment with NumPy, Pandas, Matplotlib, and Seaborn. The model iterates on code based on execution results, generating visualisations and data analysis.

Pro/Ultra bonus: Upload entire GitHub repositories or code folders up to 5,000 files.

Memory and personalisation

Memory launched August 2025, allowing Gemini to recall details and preferences across conversations.

How it works:

• Automatic recall from past conversations
• Explicit saves: “Remember my favourite colour is green”
• Manage saved memories in Settings

Temporary Chat creates ephemeral conversations:

• Deleted after 72 hours
• Won’t contribute to memory or personalisation
• Similar to ChatGPT’s Temporary Chat feature

AI Mode in Google Search

Distinct from the Gemini app, AI Mode integrates Gemini capabilities directly into Google Search for complex queries.

Key features:

• Powered by a custom Gemini model with a “query fan-out” technique
• Breaks questions into subtopics, conducts parallel searches
• Supports text, image, and voice inputs
• Conversational follow-ups

Agentic capabilities (Ultra, US only): Book restaurants, event tickets, and flights directly from search results via Project Mariner integration.

Strengths

Industry-leading context window — 1 million tokens on the Pro models (Gemini 3.5 Pro targets 2M) lets you process entire codebases, contracts or research papers in one conversation, matching or exceeding rivals.

Deep Google ecosystem integration — Built into Gmail, Drive, Docs, Calendar, Search, Android and Chrome. No competitor offers this level of native integration with products billions already use.

Strong multimodal and media — Native multimodality plus Veo and the new unified Gemini Omni for video, and Imagen / Nano Banana for images, in one app.

Capable, generous free tier — Gemini Live voice, image generation and Deep Research are all available free, with Gemini 3.5 Flash as the default model.

Value bundling — The $19.99/month Pro tier includes 2TB of Google One storage; Ultra bundles YouTube Premium — effectively a discount on the AI.

Enterprise-ready compliance — Among the first AI products certified to ISO 42001, with FedRAMP High, HIPAA, SOC 1/2/3 and comprehensive EU data-residency options.

Limitations

Still second to ChatGPT in reach — Despite 900M+ users, Gemini trails ChatGPT’s billion-plus, and the brand-recognition gap persists.

Coding isn’t best-in-class — Claude Opus 4.8 leads the hardest coding benchmark (88.6% SWE-bench Verified, 69.2% Pro) ahead of Gemini. For top-end programming, Claude remains the developer favourite.

Shipping cadence — Gemini 3.5 Pro slipped past its I/O announcement and is still in limited rollout, a reminder that the flagship isn’t always available on time.

UX acknowledged as needing work — Internal documents reportedly reference a “UX 2.0 overhaul” as a priority. The interface feels less polished than ChatGPT.

Regional restrictions frustrate users — Video generation unavailable in EEA, Switzerland, UK. Project Mariner US-only. Some features lag behind global rollout.

Conservative content policies — Gemini is more cautious than competitors on certain content types, which some users find limiting.

Who should use Gemini?

Great for:

• Google ecosystem users — If you live in Gmail, Docs, and Drive, Gemini’s extensions create genuine workflow improvements
• Long document processing — The 1M token context handles contracts, research papers, and codebases that break other assistants
• Value seekers — The free tier is remarkably capable; Pro bundles significant storage value
• Multimodal creators — Image and video generation in one platform
• Enterprise/compliance-focused teams — Best-in-class certifications including ISO 42001

Look elsewhere if:

• Coding is your primary use — Claude leads the hardest coding benchmarks, which matters for professional development
• You need mature memory — ChatGPT’s cross-conversation recall is more sophisticated
• Third-party integrations matter — ChatGPT’s 3M+ Custom GPTs ecosystem dwarfs Gems
• You’re outside Google’s ecosystem — The value proposition weakens significantly if you don’t use Google services

Historical timeline

Privacy and data handling

Understanding how Google handles your data is crucial, especially given the extensive ecosystem integration.

Training data policy

To opt out: Settings > Apps > Gemini Apps Activity > Turn off

Data retention

• Default retention: 18 months (configurable to 3–36 months)
• Human review: A subset of conversations may be reviewed for quality
• Reviewed conversations: Retained up to 3 years even after user deletion
• Temporary Chat: Deleted within 72 hours

Workspace and Enterprise privacy

Google Workspace accounts receive stronger protections:

• Customer data is NOT used for training without explicit permission
• No human review of business conversations
• Prompts and responses in Workspace side panels not retained after sessions
• Strict data isolation between customers

Security certifications

Gemini holds industry-leading compliance credentials:

• ISO 42001 — First AI product to achieve this certification
• ISO 27001/27017/27018/27701 — Information security management
• SOC 1/2/3 — Service organisation controls
• FedRAMP High — US government cloud security (July 2025)
• HIPAA — Healthcare compliance with BAA available
• GDPR — EU data residency options available
• FERPA/COPPA — Educational compliance

Visit Google’s AI Privacy Hub for detailed documentation.

Gemini vs alternatives

Quick comparison

Where Gemini wins

• Context and multimodal — 1M-token windows (2M coming) and strong native multimodality
• Free voice mode — Gemini Live is free, where some rivals gate real-time voice
• Google integration — Native access to Gmail, Drive, Docs, Search and Android that competitors can’t match
• Bundled value — 2TB storage with Pro; YouTube Premium with Ultra

Where competitors win

• ChatGPT — Larger reach, broader ecosystem and custom GPTs, agent momentum
• Claude — Leads the hardest coding benchmarks and long-form writing, with the Cowork desktop agent

Getting started

Web

1. Go to gemini.google.com
2. Sign in with your Google account
3. Start chatting—no additional setup required

Mobile

• iOS: Download from App Store
• Android: Download from Google Play

On Android, Gemini can replace Google Assistant as your default assistant.

Tips for new users

Enable Extensions: Go to Settings > Extensions and connect Gmail, Drive, Calendar for the full experience. Prompts like “What meetings do I have tomorrow?” work seamlessly.

Try Gemini Live: Tap the waveform icon for hands-free conversation. Works while walking, cooking, or whenever typing isn’t convenient.

Use the context window: Upload long documents directly. Gemini can process ~1,500 pages in a single conversation—don’t break documents into chunks.

Create a Gem: For repeated tasks, build a custom Gem with specific instructions. A “Code Reviewer” Gem with your team’s style guide saves setup time on every conversation.

Try Deep Research: For complex questions requiring multiple sources, click “Deep Research” for an autonomous agent that produces cited reports.

Frequently asked questions

Is Gemini free?

Yes. The free tier includes Gemini 3.5 Flash as the default model, limited access to the Gemini 3.1 Pro flagship, Gemini Live voice mode, basic image generation, a monthly Deep Research allowance and Google Workspace extensions. It’s genuinely capable, not just a demo.

Is Google AI Pro worth $19.99/month?

If you use Google services, yes. The bundled 2TB storage is worth $9.99/month separately, meaning you’re effectively paying $10/month for AI features including the full 1M token context, unlimited image generation, video creation with Veo, and 20 Deep Research reports daily.

What about the $99.99 Ultra tier?

Only for power users who need Gemini 3 Deep Think, heavy video generation, or Project Mariner browser automation. Google recently cut the price to $99.99/month (from $249.99). The bundled YouTube Premium ($13.99/month value) and 30TB storage help justify the premium.

How does Gemini compare to ChatGPT?

Gemini offers very large context windows (1M tokens, 2M coming on 3.5 Pro), free real-time voice and native Google integration across Search, Workspace and Android. ChatGPT has greater reach, a larger ecosystem and strong agentic momentum (Agent Mode, Codex). Benchmarks are competitive; Claude leads the hardest coding tests, while Gemini is strongest on long context and multimodal. See ChatGPT vs Claude for a related head-to-head.

Is my data private?

Free tier data is used for training by default—opt out in Settings. Workspace and Enterprise accounts have stronger protections: data is never used for training, human review doesn’t occur, and strict isolation prevents cross-customer exposure. Gemini holds ISO 42001 (first AI product certified), FedRAMP High, HIPAA, and comprehensive GDPR compliance.

Can Gemini access my Gmail and Drive?

Yes, through Extensions. Enable them in Settings to search emails, find documents, and take actions across Google services. For Workspace accounts, this requires admin approval.

Does Gemini remember previous conversations?

Yes, since August 2025. Memory automatically recalls details and preferences from past conversations. You can explicitly save information (“Remember I prefer British English”) and manage saved memories in Settings. Use Temporary Chat for private conversations that won’t be remembered.

What is Gemini Omni?

Gemini Omni is Google’s unified multimodal video model, announced at Google I/O on 19 May 2026. Unlike Veo (a text-to-video generator), Omni takes text, image, audio and video in a single prompt and returns video, edited photos or a digital avatar, with shared context across modalities. It supports conversational editing (iterate on a clip in chat), keeps characters consistent across shots, and has an improved grasp of physics. Gemini Omni Flash is rolling out to AI Plus, Pro and Ultra subscribers and free on YouTube Shorts, capped at 10-second clips for now; a higher-end Gemini Omni Pro has been teased without a date.

What happened with the image generation controversy?

In February 2024, Gemini’s image generator produced historically inaccurate images due to overzealous diversity prompting (generating diverse Nazi soldiers, female popes). Google paused people generation for six months and publicly apologised. Image generation returned in August 2024 with improved guardrails.

Verdict

Gemini has gone from Google’s ChatGPT panic response to a genuinely competitive, frontier-class assistant — now with 900 million-plus monthly users. Its strengths are very large context windows, strong native multimodality (Veo, Imagen), and being woven into Search, Android, Chrome and Workspace, backed by Google’s full-stack advantage of owning its own TPUs and cloud.

Free tier: Genuinely useful — Gemini 3.5 Flash as default, free real-time voice, image generation, Deep Research and Workspace extensions.

Google AI Pro ($19.99/month): The sweet spot for Google users. The bundled 2TB of storage effectively halves the cost of the AI; you get Gemini 3.1 Pro, a 1M-token context, Veo video and higher Deep Research limits.

Google AI Ultra ($99.99/month): For power users who need Deep Think reasoning, early Gemini 3.5 Pro access, heavy video generation or Project Mariner automation, with YouTube Premium bundled in.

Gemini isn’t best at everything — Claude leads the hardest coding benchmarks and long-form writing, and ChatGPT has greater reach and agent momentum — but its context length, multimodal breadth and Google integration are advantages neither fully matches. For anyone in Google’s ecosystem, it’s increasingly the default.

Official links