ChatGPT vs Claude vs Gemini

ChatGPT, Claude and Gemini all cost about $20 a month, so the choice comes down to what you do most. ChatGPT Plus is the broadest: images, voice, deep research, a coding agent and unlimited everyday chat in one app. Claude Pro gives the best coding, and writing level with Gemini's best, at that price, but it makes no images or video and measures usage in 5-hour sessions. Gemini through Google AI Pro is the value pick for Google users, with Gemini inside Gmail and Docs, the only built-in video generator of the three, and 5 TB of storage.

Which one should you pay for?

Pay for the one that matches your main job. If you can't name one, ChatGPT Plus is the safest default because it does the most things reasonably well.

• Mixed everyday use (questions, email, planning, the odd image): ChatGPT Plus. Everyday text chat is unlimited on every ChatGPT plan, subject to abuse guardrails, and Plus adds the thinking models, deep research, Codex and image creation with thinking. Whether $20 is worth it over the free plan is its own question, covered in Is ChatGPT Plus worth it?
• Writing and editing: Claude Pro. Claude Opus 5.5 and Gemini 4 Argon share the top of LMArena's creative-writing ranking, well ahead of every OpenAI model, and Claude Pro includes Opus 5.5.
• Coding: Claude Pro, for Claude Code with Opus 5.5, first in LMArena's WebDev ranking. ChatGPT Plus with Codex is a close second.
• Research and web search: ChatGPT Plus. OpenAI models hold the top three places in LMArena's Search Arena, with style control on.
• Images: ChatGPT. OpenAI's Images 2.5 models are first and second in LMArena's text-to-image ranking, and Claude makes no images at all.
• Video: Gemini. Gemini Omni is the only video generator built into any of the three apps, from the $4.99 Google AI Plus plan up.
• Gmail, Docs and Drive users: Gemini with Google AI Pro ($19.99), which puts Gemini inside Gmail and Docs and includes 5 TB of storage.
• Students: Gemini, because eligible US college students can get a year of Google AI Pro free. ChatGPT's student offer gives four months of Plus if claimed by October 31, 2026. Claude is for users 18 and older only.
• Tight budget: Google AI Plus at $4.99 a month or ChatGPT Go at $8. Claude has nothing between free and $20.
• Heavy users who hit limits: all three sell $100 and $200 tiers. Claude Max is the one to pick for all-day coding; ChatGPT Pro for all-day research and voice.

Skip ChatGPT Plus if you mostly write or code and never make images. Skip Claude Pro if you want images, video or long voice chats, or if running out of messages mid-task would annoy you. Skip Google AI Pro if you don't use Google's apps and do a lot of coding.

How much do ChatGPT, Claude and Gemini cost?

The main plans cost the same, $20 a month for ChatGPT Plus and Claude Pro and $19.99 for Google AI Pro, as of October 2026. All three also sell tiers at about $100 and $200. They differ at the bottom: Google and OpenAI sell a cheap plan, Anthropic doesn't.

Tier	ChatGPT	Claude	Gemini (Google AI plans)
Free	Free, may show ads	Free	Free, 15 GB storage
Budget	Go, $8 per month, may show ads	None	AI Plus, $4.99 per month, 400 GB storage
Main	Plus, $20 per month (monthly only)	Pro, $20 per month, or $17 per month billed yearly ($200)	AI Pro, $19.99 per month, 5 TB storage
About $100	Pro 100, $100 per month; 15 hours of voice a day	Max 5x, $100 per month: 5x Pro per session, plus $100 a month in API credits	AI Ultra, $99.99 per month: 5x AI Pro, 20 TB storage, YouTube Premium
About $200	Pro 200, $200 per month; more usage than Pro 100, unlimited voice	Max 20x, $200 per month: 20x Pro per session, plus $200 a month in API credits	AI Ultra, $199.99 per month: 20x AI Pro
Top	Pro 500, $500 per month: the highest allowance, plus Ultrafast mode	None	None

Sources: OpenAI's Plus, Pro tiers and voice help pages, OpenAI's Go announcement for the $8 US price, Anthropic's pricing page and Max plan page, and Google's plans page. OpenAI's pricing page shows local prices outside the US.

None of the three publishes message counts for its paid plans. ChatGPT advertises unlimited everyday text chat on every plan, subject to abuse guardrails (Plus's help page still reserves the right to cap messages at busy times); Claude and Gemini measure usage in 5-hour windows with a weekly ceiling, and both Claude Pro and Max can buy usage credits to keep going.

Three things the sticker prices hide:

• Google's plans are Google One storage plans too, so AI Pro replaces whatever you pay Google for storage today.
• ChatGPT's Free and Go plans can show ads. OpenAI started testing ads in the US on February 9, 2026, and its ads FAQ says Plus, Pro, Business, Enterprise and Edu accounts will not have them.
• ChatGPT Pro 200 is getting smaller. New Pro 200 subscriptions already get a lower allowance, and existing subscribers keep the old one only through October 29, 2026, then move to the lower allowance at the same $200, per OpenAI's Pro tiers page.

Are the free versions of ChatGPT, Claude and Gemini good enough?

Yes, for casual use. As of October 2026, ChatGPT Free has unlimited everyday text chat, subject to abuse guardrails, Gemini Free includes the most features, and Claude Free has the strongest default free model but the tightest limits. All three free plans run lighter models than the paid ones.

Free plan	ChatGPT Free	Claude Free	Gemini Free
Model	Luna, OpenAI's light model	Sonnet and Haiku (currently 5.5)	Flash-Lite, or Auto, which can route harder prompts to Flash or Pro
Chat limit	Unlimited everyday text chats, subject to abuse guardrails	Session limit that resets every 5 hours	Compute-based, refreshes every 5 hours, weekly cap
File uploads	3 files per day	Allowed, no number published	Up to 10 files per prompt, rolling limits
Context window	27K tokens in Instant mode (about 12 pages of input)	Up to 1M tokens, varies by model	32K tokens
Image creation	Limited, no number published	None	Yes; on the default Flash-Lite model, Nano Banana 2 Lite (downloads at 1K)
Video creation	None	None	None in the Gemini app (needs a paid plan)
Deep research	Limited	No (Research is a Pro feature)	Yes
Voice	Limited, on the smaller GPT-Live-1.1 Quip	Voice mode	Gemini Live
Ads	May appear	None mentioned	None mentioned

Sources: OpenAI's pricing page, free tier FAQ, upload limits and voice limits; Anthropic's pricing page and getting started page; Google's limits page, models page, upload page and image page.

Which Luna model ChatGPT Free runs depends on which OpenAI page you read. The release notes say Free and Go moved to GPT-6 Luna from October 8, 2026; the models and free tier help pages still name GPT-5.6 Luna. Either way it is OpenAI's light model, not the GPT-6 Sol that Plus gets.

Which free plan to use:

• ChatGPT Free if you chat a lot and rarely attach files. The three-uploads-a-day limit is the wall most people hit first.
• Gemini Free if you want tools without paying: Deep Research, image creation, Gemini Live and Canvas are all free, and the free plan includes limited Google Flow access, with Google's top Nano Banana Pro image model. The 32K-token context window is small, so long documents get cut short.
• Claude Free for the best free writing and analysis. At the same Medium effort, Claude Sonnet 5.5 scores 41 on the Artificial Analysis Intelligence Index against 30 for GPT-6 Luna, and it leads on LMArena's text ranking too (1477 vs 1441). The catch is the 5-hour session limit, which long chats and big attachments use up fast.

What do you get for $20 a month?

At $20, ChatGPT Plus gives the most features, Claude Pro gives the strongest model for code and one level with Gemini's for writing, and Google AI Pro gives the most extras.

	ChatGPT Plus	Claude Pro	Google AI Pro
Price	$20 per month, monthly only	$20 per month, or $17 billed yearly	$19.99 per month
Best model in chat	GPT-6 (Sol) at Medium and High thinking, still rolling out; GPT-5.6 Sol selectable	Claude Opus 5.5; Fable 5 and 5.1 only on paid usage credits	Gemini 4 Argon (rolling out from October 13, 2026)
Usage	Unlimited everyday chat; separate limits for thinking and tools	At least 5x Free per 5-hour session, plus a weekly limit	4x Free, refreshes every 5 hours, weekly limit
Context window	54K tokens Instant, 256K tokens with reasoning	Up to 1M tokens, varies by model	1M tokens (about 1,500 pages)
Images	ChatGPT Images 2.5, including images with thinking	None	Nano Banana 2, with redo in Nano Banana Pro
Video	None	None	Gemini Omni, 1,000 Google Flow credits
Coding agent	Codex, with GPT-6 Astra and GPT-6.1 Sol	Claude Code	Antigravity and Jules, at higher limits than free
Voice	Up to 3 hours a day with GPT-Live-1.1	Voice mode	Gemini Live
Extras	No ads, Sites, add-ins for Word, Excel, PowerPoint and Google Sheets	Claude in Chrome, Claude for Microsoft 365, Claude for Google Workspace (beta)	5 TB storage, Gemini in Gmail and Docs, YouTube Premium Lite

Sources: OpenAI's pricing page and models page; Anthropic's pricing page, Pro plan page and Fable page; Google's plans page, limits page and release notes.

Two caveats on the model row:

• ChatGPT: Plus runs GPT-6 Sol in chat, and OpenAI's models page says GPT-6 is still rolling out, so some Plus accounts don't see it yet. OpenAI's stronger GPT-6 Astra reaches Plus only inside ChatGPT Work and Codex; in chat, Astra is a Pro-plan model.
• Claude: Anthropic's Fable models (Fable 5 and 5.1) sit above Opus in some rankings, but on Pro they run only on pay-as-you-go usage credits. On Max they are part of the plan, for up to 50% of the weekly limit.

Which is smartest right now?

On independent rankings in October 2026, the flagship models from Google and Anthropic lead the chat rankings, and OpenAI leads only in search and images. Gemini 4 Argon is first in LMArena's text ranking, Claude Opus 5.5 is third, and the best OpenAI model is 21st. Take Argon's lead as provisional: it launched on October 13 and has fewer than 5,000 votes.

Ranking (date)	Best Gemini model	Best Claude model	Best OpenAI model
LMArena Text, all prompts (Oct 20)	4 Argon: 1st, 1524	Opus 5.5: 3rd, 1509	GPT-5.6 Sol: 21st, 1484
LMArena Text, creative writing (Oct 20)	4 Argon: 2nd, 1516	Opus 5.5: 1st, 1520	GPT-5.6 Sol: 19th, 1469
LMArena WebDev, building web apps (Oct 19)	4 Argon: 10th, 1678	Opus 5.5: 1st, 1807	GPT-6 Astra: 2nd, 1774
LMArena Search (Oct 13)	3.1 Pro: 14th, 1207 (4 Argon not yet ranked)	Fable 5.1: 4th, 1233	GPT-5.6 Sol: 1st, 1258
LMArena Text to Image (Oct 7)	Nano Banana 2.1: 6th, 1328	No image model	GPT-Image-2.5 Sunburst: 1st, 1425
LMArena Text to Video (Sep 22)	Gemini Omni 1.1 Flash: 1st, 1516	No video model	Sora 2 Pro: 11th, 1368 (discontinued)
Artificial Analysis Intelligence Index (Oct 22)	4 Argon (high): 53	Opus 5.5 (max): 58	GPT-6 Astra (max): 53

Sources: LMArena's public leaderboard dataset (style-controlled scores where LMArena offers them, for Text and Search; the other arenas have one score; published on the dates shown) and the Artificial Analysis leaderboard. LMArena ranks models by blind votes from people comparing two answers; Artificial Analysis averages a set of tests it runs itself. Vendor benchmarks are left out because each maker picks its own tests.

Three things to know before reading too much into this:

• The apps don't run the ranked settings. These rankings test the models through their APIs, often at the highest effort setting, while the apps choose their own; Anthropic says its apps default to Medium effort.
• Plus's chat default ranks lower than its older model. GPT-6 Sol ranks 62nd in LMArena's text ranking, below the older GPT-5.6 Sol that Plus users can still pick.
• The gaps between the top models are small. On LMArena's scale, a 15-point lead means the leader is expected to win about 52 percent of head-to-head votes.

Which is best for writing?

Claude and Gemini are the best of the three for writing, as of October 2026, and both are well ahead of ChatGPT. In LMArena's creative-writing ranking (October 20), Claude Opus 5.5 scores 1520 and Gemini 4 Argon 1516: a tie, on 2,241 and 1,055 votes. OpenAI's best model there places 19th.

For writers the choice is Claude Pro, which has the longer track record and Claude Free as a strong free fallback, or Google AI Pro if you also want Gemini inside Google Docs. ChatGPT is the weakest of the three on this ranking.

Which is best for coding?

Claude Pro is the best $20 plan for coding, with ChatGPT Plus a close second. The evidence is LMArena's WebDev ranking, where people vote on working web apps: Claude Opus 5.5 is first (1807) and OpenAI's GPT-6 Astra second (1774), and their confidence ranges don't overlap (October 19, 2026). LMArena's Agent Arena, which scores models on real agent sessions, agrees: Opus 5.5 leads Astra, 0.21 to 0.15 (October 20). In the coding category of the plain text ranking, which scores code answers in chat, Gemini 4 Argon is first and Opus 5.5 fifth.

• Claude Code comes with Claude Pro ($20 a month) and isn't on the free plan. Long coding sessions run into Pro's 5-hour session limit; Max at $100 or $200 a month raises it 5x or 20x.
• OpenAI Codex comes with every ChatGPT plan: limited on Free and Go, fuller on Plus with GPT-6 Astra and GPT-6.1 Sol. Codex and ChatGPT Work share one allowance, and weekly limits may apply on top of the five-hour ones, per OpenAI's Codex pricing page.
• Google Antigravity, Google's coding agent, is free for individuals and built into the Gemini Mac app since October 13. Google AI Pro raises its limits. Gemini 4 Argon ranks 10th in WebDev, so Gemini is the third pick for code.

Which is best for research and web search?

ChatGPT is the best of the three for web research. OpenAI models take the top three places in LMArena's Search Arena with style control on (October 13, 2026), led by GPT-5.6 Sol at its xHigh setting. ChatGPT Plus users can select GPT-5.6 Sol, but Plus chat thinks only at Medium and High, so the ranked setups are not exactly what Plus runs. Plus also has "expanded" deep research where Free and Go are "limited", per OpenAI's pricing page.

Gemini is the better free choice for research reports: Deep Research is on Gemini's free plan, ChatGPT Free's deep research is limited, and Claude keeps its Research feature for paid plans. For long uploaded sources, Google AI Pro's 1M-token context window holds about 1,500 pages, against 256K tokens for ChatGPT Plus with reasoning.

Which is best for images and video?

ChatGPT is best for images and Gemini is the only choice for video. Claude makes neither: Anthropic's help page says Claude "doesn't generate photos or illustrations", though it can draw charts and diagrams in code.

• Images: OpenAI's Images 2.5, in ChatGPT since September 2026, is available on every plan, with "images with thinking" on Plus and Pro. The API versions of Images 2.5 rank first and second in LMArena's text-to-image ranking (October 7). Gemini's free images come from Nano Banana 2 Lite on the default Flash-Lite model, and paid plans get Nano Banana 2 with a redo in Nano Banana Pro, per Google's image help page.
• Video: Gemini Omni makes and edits videos in the Gemini app on any Google AI plan for users 18 and older, per Google's video help page, and Gemini Omni models are first and second in LMArena's text-to-video ranking (September 22). ChatGPT no longer makes video: OpenAI discontinued the Sora app on April 26, 2026. Google AI Plus at $4.99 a month is the cheapest way in, with 200 Google Flow credits. The best free AI video generators guide covers the free routes.

Which is best if you use Gmail, Docs and Drive?

Gemini, through Google AI Pro at $19.99 a month. It is the only one of the three that works inside Gmail itself, and AI Pro adds Gemini in Docs, 5 TB of Google storage and Chrome's auto browse, per Google's plans page. The $4.99 AI Plus plan includes Gemini in Gmail but not Docs. If your Google account comes from work or school, Gemini's features depend on your organization's Workspace license instead of these plans.

ChatGPT and Claude now reach into Google's files too, so pick them for the model rather than ruling them out. Claude for Google Workspace, in beta, adds a Claude sidebar inside Docs, Sheets and Slides that edits the open file (Anthropic help), and Claude's connectors can search Drive, Gmail and Calendar from a chat. OpenAI's release notes say ChatGPT for Excel and Google Sheets is available globally, alongside its Word and PowerPoint add-ins. For Microsoft's apps, Claude Pro includes Claude for Microsoft 365.

Which is best for students?

Gemini is the best deal for students. Eligible US college students can get a year of Google AI Pro at no cost, and students in more than 140 other countries a year of AI Plus, if they redeem by December 31, 2026, per Google's release notes. Gemini's free plan also has quizzes, flashcards and Deep Research.

ChatGPT has a shorter offer: four free monthly billing periods of Plus for US college and graduate students who claim it by October 31, 2026, verified through SheerID, then $20 a month unless you cancel (OpenAI help). Subscriptions billed through Apple or Google Play can't take it. After that date, ChatGPT's free plan, with Study Mode on every plan, is the fallback.

Claude is the weakest fit for students: you must be 18 or older to use it, per Anthropic's getting started page, and Anthropic says it offers no standard discount on Pro or Max. Your university may already provide one of them through ChatGPT Edu, Claude's Education plan or Google's education licenses, so check before paying.

What do AI assistants get wrong about this?

Asked this question on October 22, 2026, Perplexity's Sonar recommended "Gemini Advanced", a plan name Google no longer uses, said it could not find current prices for Claude or Gemini, and said ChatGPT's free plan "will feel constrained" with heavy use, when everyday text chat is unlimited on every plan, subject to abuse guardrails. Asked again by a reviewer the same day, Sonar gave the same broad picks with no prices or free-plan limits, so answers vary from run to run. That is why this guide links each price and limit to the maker's page and dates it.

What is the same in all three?

Each runs on the web, iOS, Android and desktop apps for Mac and Windows, searches the web, reads PDFs, spreadsheets and images, and keeps related files and chats together (projects in ChatGPT and Claude, notebooks in Gemini). All three change fast: each shipped a new flagship model between September and October 2026, so expect the rankings to move.

How to decide

Start on the free plans and give each one the same three real tasks from your week. Pay for the one you keep going back to. If the tasks are writing or code, start your trial with Claude; if they involve Gmail, Docs or video, start with Gemini; for everything else, start with ChatGPT. Pay monthly for the first month (only Claude offers a yearly discount, and only on Pro), and move up to a $100 tier only if you hit the limit on most working days.