ChatGPT vs Claude vs Gemini in 2026: differences and which to choose

ChatGPT, Claude and Gemini are the three best-known AI assistants today, and all three do many similar things: they answer questions, write, summarise documents, search the web and help with code. The real differences are in the details: what each plan includes, which models each service uses, what tools surround it and which other apps it connects to.

This comparison doesn't declare a winner. It explains what each one offers, what it costs and which kinds of tasks it may suit best, based on official documentation from OpenAI, Anthropic and Google. We focus on the consumer products (the ChatGPT, Claude and Gemini apps and websites), not on their developer APIs or business plans.

ChatGPT, Claude and Gemini in a nutshell

• ChatGPT (OpenAI) is the most general-purpose of the three: chat, voice, image generation, web research and, since 2026, a desktop app that combines chat with two more modes: Work, for handing off long tasks, and Codex, for coding.
• Claude (Anthropic) focuses on text, document analysis and coding. It can create files, run code and work with connected apps, but it doesn't generate photos or illustrations.
• Gemini (Google) stands out for its integration with Google's ecosystem: Gmail, Docs, Drive, Chrome and Android. Its paid plans also include Google One storage and video generation tools.

All three have a free plan with usage limits and several paid plans. None is better at everything: the choice depends on what you'll use it for and which tools you already use.

Main differences between ChatGPT, Claude and Gemini

• Image generation: ChatGPT and Gemini create and edit images, including on their free plans. Claude analyses images but doesn't generate photos or illustrations; it can create diagrams and charts.
• Ecosystem: Gemini is built into Google's apps. ChatGPT and Claude connect to outside services (Gmail, Google Drive, Microsoft 365 and others) through integrations the user turns on.
• Coding: all three ecosystems have dedicated tools beyond the chat: Codex for ChatGPT, Claude Code for Claude and Google Antigravity (with Jules) for Google.
• Research: ChatGPT and Gemini offer their in-depth research mode on the free plan, with limits; in Claude, the Research feature requires a paid plan.
• Entry price: ChatGPT (Go) and Gemini (Google AI Plus) have mid-tier plans under 10 dollars a month. Claude's first paid plan is Pro, at 20 dollars a month.
• Extras beyond AI: Google's plans include cloud storage (from 400 GB) and, depending on the plan, other Google services. ChatGPT and Claude don't include services outside the assistant.

Plans and prices

Prices in this section are the list prices each company publishes in US dollars, unless stated otherwise. The final price may change depending on the country, taxes, currency and whether you subscribe on the web or through a mobile app store.

ChatGPT plans

• Free ($0): chat with limits, image generation, voice, web search, limited deep research, basic memory and projects with few files.
• Go ($8/month in the US): more messages, file uploads and images than the free plan. OpenAI adjusts its price in some markets and offers it in every country where ChatGPT is available.
• Plus ($20/month): advanced reasoning models, expanded limits, more deep research, more memory and context, projects, scheduled tasks and custom GPTs.
• Pro ($100, $200 or $500/month): three tiers that differ mainly in how much usage they include. Pro 100 and Pro 200 include the GPT-6 Pro model; Pro 200 is still open to new subscribers, with less included usage than before. Pro 500, announced on 29 September 2026, has the most usage and is the only one of the three with Astra Ultrafast, a faster mode for Work and Codex.

Claude plans

• Free ($0): chat on web, desktop and mobile, web search, file creation and code execution, memory, connectors, voice mode and up to 5 projects.
• Pro ($20/month, or $17/month with a $200 annual payment): more usage, more models (including Opus), Research, Claude Code, Claude in Chrome, Claude for Microsoft 365, projects without the 5-project limit and the ability to hand off and schedule tasks.
• Max ($100 or $200/month): 5 or 20 times Pro's usage, higher output limits, early access to features and priority at high-traffic times. Billed monthly only.

According to Anthropic, Claude's prices don't include tax and may differ if you subscribe through a mobile app store.

Gemini plans

• Free: access to the Gemini 3.6 Flash model, varying access to 3.1 Pro, image generation and editing, Deep Research, Gemini Live, Canvas and 15 GB of storage.
• Google AI Plus ($4.99/month in the US; €4.99/month in Spain): twice the free plan's usage, video generation, Gemini in Gmail and other Google apps, and 400 GB of storage.
• Google AI Pro ($19.99/month in the US; €21.99/month in Spain): four times the free plan's usage, more access to 3.1 Pro, a 1 million token context window, Gemini in Gmail and Docs, coding tools and 5 TB of storage.
• Google AI Ultra (from $99.99/month in the US; from €99.99/month in Spain): two tiers, with 5 or 20 times AI Pro's usage (the 20x tier costs $199.99 in the US and €219.99 in Spain), early access to features such as Deep Think and at least 20 TB of storage.

Google publishes different prices by country: in Spain the AI Pro plan costs €21.99 a month and the top Ultra tier €219.99, compared with $19.99 and $199.99 in the United States. Benefits also vary by country; for example, the Spanish page doesn't include every feature listed on the US one.

Models available in 2026

It's worth separating the product from the model. ChatGPT, Claude and Gemini are the apps; inside each one run one or more language models that the company changes and updates over time, often without the user having to do anything. As of 3 October 2026, according to official documentation:

• ChatGPT: in chat, Free and Go use GPT-5.6 Luna by default; Plus and Pro use GPT-5.6 Sol, with reasoning levels from Instant to Extra High, and Pro adds the GPT-5.6 Sol Pro option. OpenAI’s current Chat Help lists GPT-6 Pro, powered by Astra, for Pro $100 and $200, Business and Enterprise; Plus includes Astra in Work and Codex, but not GPT-6 Pro in Chat. The Work and Codex modes use GPT-6 generation models that aren't in chat (GPT-6.1 Sol, GPT-6 Sol and GPT-6 Luna) and, from Plus, GPT-6 Astra; on Free and Go, Codex offers GPT-6 Luna in the desktop app, depending on the rollout.
• Claude: the current family is Haiku 4.5, Sonnet 5.5, Opus 5.5 and Fable 5.1. The free plan includes Haiku and Sonnet; paid plans add Opus and Fable, the latter with usage capped at a share of the weekly allowance. The model is chosen in a picker, together with the reasoning level.
• Gemini: the free plan uses Gemini 3.6 Flash and has varying access to 3.1 Pro; paid plans expand access to 3.1 Pro, and Ultra gives early access to Deep Think. The model is chosen from a drop-down menu.

Model names change several times a year, and the companies describe each new version as their most capable. When choosing an assistant, the plan, limits and tools matter more than the exact model name. For a closer look at GPT-6 Astra’s capabilities and examples that identify it by name, read our GPT-6 Astra guide.

ChatGPT vs Claude vs Gemini for writing

All three draft, summarise, translate and rewrite text fluently. Differences in style exist, but they're subjective and change with every model update, so there's no reliable answer to which one writes best.

• ChatGPT offers the widest range of tools around text: projects, custom GPTs, memory and images in the same conversation.
• Claude can create and edit documents, spreadsheets and presentations as downloadable files, and its paid plans include Claude Docs and Slides.
• Gemini brings its writing help straight into Gmail and Google Docs on paid plans (in selected languages), as well as Canvas in the app.

The most practical approach is to try the same task on all three free plans and keep the style you have to correct least.

Which is best for coding

There's no single answer. All three generate, explain and fix code in the chat, and all three also have dedicated developer tools. Coding benchmark results are mostly published by the companies themselves, with different versions and methods, so we don't use them to declare a winner.

• Codex (ChatGPT): OpenAI's coding agent works in the ChatGPT desktop app, the terminal and the IDE. It's included on all plans, Free and Go too, which use it with GPT-6 Luna in the desktop app. Codex Cloud, the version that works in the cloud, isn't on Free or Go: it requires an eligible Plus account or any Pro tier.
• Claude Code (Claude): Anthropic's agent works in the terminal, the IDE, the desktop app and the web. It's included on all paid plans and shares usage limits with the chat.
• Google Antigravity and Jules (Gemini): in 2026 Google replaced its Gemini CLI terminal tool with Antigravity CLI and the Antigravity platform. Jules, its asynchronous agent, is still in beta and only officially supports English.

If you already work in one of these ecosystems, the simplest option is usually to start with its coding tool. If not, all three let you try the entry-level plan before paying more.

Research and web search

Two things are worth telling apart: quick web search, where the assistant checks a few sources and answers with links, and in-depth research, where it works for several minutes, consults many sources and delivers a report.

• ChatGPT: web search on all plans and deep research with limits that depend on the plan, including the free one.
• Claude: web search on all plans, including the free one. Research, its multi-step research mode, is available on Pro and Max.
• Gemini: Deep Research is included from the free plan, with limits, and can draw on Google Search.

In all three cases, answers include the sources consulted, but it's worth opening and checking them: assistants can summarise badly or cite outdated information.

Files, PDFs and long context

All three accept PDFs, documents, spreadsheets and images, with different limits:

• ChatGPT: up to 512 MB per file as the hard limit; in addition, each text document is capped at 2 million tokens, each image at 20 MB and spreadsheets at about 50 MB. The free plan allows far fewer uploads than the paid ones.
• Claude: up to 500 MB per file and 20 files per conversation; in projects, 30 MB per file.
• Gemini: Google mentions up to 1,500 pages of documents and code repositories of up to 30,000 lines on its paid plans.

The context window is the amount of information the model can take into account at once. Anthropic lists a context of up to 1 million tokens on its Claude plans, depending on the model, and Google lists it for Google AI Pro. OpenAI doesn't publish an equivalent figure for the ChatGPT app, and developer API limits aren't necessarily those of the consumer product.

Images, voice and multimodal features

• Analysing images: all three understand photos, screenshots and charts you upload to the conversation.
• Generating images: ChatGPT does it with ChatGPT Images 2.0, on all plans. Gemini generates and edits images from the free plan and also has Google Flow. Claude doesn't generate photos or illustrations; it creates diagrams, charts and visualisations with code.
• Video: Gemini includes video generation on its paid plans, with different levels of access depending on the plan. Claude doesn't generate video.
• Voice: ChatGPT Voice's Live mode uses GPT-Live-1 on paid plans and GPT-Live-1 mini on Free; Gemini Live is available for free; Claude has a voice mode in beta on all plans.
• Camera and screen: Gemini Live lets you share your camera and screen for free on Android and iOS. In ChatGPT, Live mode doesn't currently support video or screen sharing; eligible subscribers can use them in the iOS and Android apps with Advanced, the previous voice experience. We haven't found an equivalent feature in Claude's documentation.

Memory, projects and integrations

• Memory: all three can remember preferences and previous conversations if you allow it. ChatGPT offers a lighter version of memory on the free plan; Claude includes memory on all its plans; Gemini now calls its feature for remembering past chats memory and lets you import memories and history from other assistants.
• Projects: ChatGPT and Claude organise chats and files into projects (Claude limits its free plan to 5). In Gemini, Gems and the new Skills feature store reusable instructions.
• Integrations: Gemini has an advantage if you live in Google: it works inside Gmail, Docs and Drive and, with Personal Intelligence, can use data from Gmail, Photos or your search history if you give it access. ChatGPT and Claude connect to services such as Google Drive, Gmail, Microsoft 365 or Slack through apps or connectors, with restrictions depending on the plan and region.

Agents and long tasks

In 2026 all three ecosystems moved from chat to agents, which carry out multi-step tasks on their own, but availability is very uneven:

• ChatGPT Work, launched in July 2026, takes a goal and works for hours with files and connected apps to deliver documents, spreadsheets or presentations. It has been rolling out across plans starting with Pro.
• Claude has merged chat with Cowork, its task mode, on Pro and Max: you can hand it a whole job that keeps running in the cloud with your laptop closed, schedule recurring tasks and, in beta, let it use your computer.
• Gemini offers its Agent Mode on the web only to Ultra subscribers in the United States, and Gemini in Chrome's auto browse to Pro and Ultra subscribers in the United States.

Agent workloads can add to the demand for inference capacity. For a closer look at the data centres, power and investment behind that infrastructure, read our explainer on the AI data-centre race.

Privacy and data controls

All three let you limit how your conversations are used, but with different rules. This is what each company's documentation says for personal accounts; it isn't a legal assessment or a ranking of which is more private.

• ChatGPT: if you turn off the model improvement setting ("Improve the model for everyone"), your new conversations aren't used for training. Temporary chats don't appear in your history and aren't used to improve the models, although OpenAI may keep a copy for up to 30 days for safety.
• Claude: on Free, Pro and Max, the setting for helping to improve Claude decides whether your chats are used for training. If it's on, Anthropic may keep the data for up to 5 years; if not, the general 30-day period applies. Conversations flagged by its safety systems may still be used for safety purposes.
• Gemini: with activity turned on, some conversations may be reviewed by people to improve Google's services, and reviewed ones are kept for up to three years even if you delete your activity. Temporary chats aren't used for training and are kept for up to 72 hours.

Business and education accounts for all three services have different terms. With sensitive data, always check your account's privacy settings.

Which to choose for what you need

These recommendations are based on verified features, not on any supposed superior quality of a model:

• General use and range of features: ChatGPT brings text, images, voice, research and agents together in one app, with a cheap mid-tier plan (Go).
• Writing: all three are competitive; choose by style and where you write (Claude for documents as files, Gemini if you write in Google Docs or Gmail).
• Coding: it depends on your workflow; Codex, Claude Code and Antigravity are extensions of each ecosystem, and Claude Code requires a paid plan.
• Research: ChatGPT and Gemini include in-depth research on the free plan; in Claude, Research is paid.
• Studying: Gemini and ChatGPT have cheap entry plans and free in-depth research; Google also runs student offers in some countries.
• Long documents: Claude and Gemini (AI Pro) publish context windows of up to 1 million tokens; ChatGPT accepts very large files but doesn't publish an equivalent context limit.
• Google ecosystem: Gemini, for its integration with Gmail, Docs, Drive, Chrome and Android.
• Creating images: ChatGPT or Gemini; Claude doesn't generate images.
• Voice and camera: Gemini Live (free, with camera and screen sharing on mobile) or ChatGPT Voice.
• Long projects and delegated tasks: ChatGPT Work or Claude on paid plans; Gemini's agent features are limited to the United States for now.

Frequently asked questions

Which is better: ChatGPT, Claude or Gemini?

None is better at everything. ChatGPT is the most general-purpose, Claude focuses on documents, files and coding, and Gemini on integration with Google. The most reliable approach is to try the free plans with your own tasks.

Which has the best free plan?

It depends on what you need. All three include web search and memory. ChatGPT and Gemini add image generation and limited in-depth research; Gemini adds Gemini Live with camera and 15 GB of storage; Claude includes connectors, file creation and up to 5 projects.

How much do the paid versions cost?

The most common paid plan costs around 20 dollars a month for all three: ChatGPT Plus ($20), Claude Pro ($20, or $17 a month paid annually) and Google AI Pro ($19.99 in the US or €21.99 in Spain). ChatGPT Go ($8 in the US) and Google AI Plus ($4.99 or €4.99) are cheaper options.

Can Claude create images?

It doesn't generate photos or illustrations. It can analyse images you upload and create diagrams, charts and visualisations with code. For image generation, ChatGPT and Gemini do have that feature.

Do they use my conversations to train their models?

It can happen with all three if you don't change your settings. Each lets you turn it off in its privacy settings and offers temporary or incognito chats that aren't used for training, with different retention periods.