In 60 seconds: ChatGPT, Claude, and Gemini all want your $20 a month — but they aren't selling you the same thing. Here's the side-by-side on price, models, limits, and context windows, with a straight recommendation for each kind of user.
- ChatGPT — Free $0, Go $8, Plus $20, Pro up to $500/month. GPT-6 models with a 1.05-million-token context window; everyday chat is unlimited on every plan, but heavy reasoning and agent work are metered.
- Claude — Free $0, Pro $20 ($17 on annual), Max $100/$200/month. Opus/Sonnet/Haiku 5.5 models, 200K-token context; limits are compute-based with no fixed message counts; Claude Code ships on paid plans.
- Gemini — Free $0, AI Plus $4.99, AI Pro $19.99, Ultra up to $199.99/month. The cheapest paid entry in the business; 1M-token context and video-generation credits on paid tiers; deepest Google-app integration.
- The $20-a-month sweet spot exists on all three — but which $20 is right depends on what you actually do with the thing.
- Every figure below comes from the vendors' own pricing pages, verified Oct 21–23, 2026 and re-scanned for announced price changes on Oct 24, 2026.
Picking an AI subscription in 2026 is a genuinely annoying decision. Three companies, three different metering systems, three different definitions of "unlimited" — and every review site ranks them by whoever bought the ad slot. So we did the boring work instead: three separate tier-by-tier explainers, each checked line by line against the official pricing pages, and now this — the honest side-by-side. No sponsored anything.
What happened
The three flagship AI assistants — OpenAI's ChatGPT, Anthropic's Claude, and Google's Gemini — each converged on roughly $20 a month as the "serious user" price. Underneath that coincidence, though, the products are barely comparable. OpenAI meters reasoning and agent work. Anthropic meters compute. Google bundles in YouTube Premium and 30 terabytes of storage at the top end. If you buy blind, you can easily pay for allowances you'll never touch — or hit a wall on the free tier doing work a $5 plan would have covered.
Here's the full comparison, built from our verified deep-dives on each service.
Latest verified facts
The short version first: every major claim in this table was verified against the vendor's own pricing page between October 21 and 23, 2026, and we re-scanned for announced price changes this morning. Prices are USD, per month, billed monthly.
| Feature | ChatGPT (OpenAI) | Claude (Anthropic) | Gemini (Google) |
|---|---|---|---|
| Cheapest paid tier | Go — $8/mo (may include ads) | Pro — $20/mo ($17/mo annual) | AI Plus — $4.99/mo |
| Mainstream tier | Plus — $20/mo | Pro — $20/mo | AI Pro — $19.99/mo |
| Top tier | Pro — $100 / $200 / $500/mo | Max 5x / 20x — $100 / $200/mo | Ultra 5x / 20x — $99.99 / $199.99/mo |
| Models you get | GPT-6 family (GPT-6, GPT-6 Astra, GPT-6.1 Sol, GPT-6 Luna) | Opus 5.5, Sonnet 5.5, Haiku 5.5 (paid); Sonnet/Haiku on Free | Gemini Pro and Flash models; Deep Think reasoning on Ultra |
| Context window | 1.05M tokens (128K max output) — shared with tools and memory | 200K tokens on personal plans | ~32K free / 128K AI Plus / 1M AI Pro and Ultra |
| What "unlimited" really means | Everyday chat unlimited on all plans; reasoning and agent (Work/Codex) work metered in 5-hour windows plus weekly caps | Compute-based — no published message counts; paid plans buy bigger usage multiples, overflow via API-rate credits | Compute-based; AI Pro gets roughly 4x free-tier usage, Ultra 5x–20x of Pro |
| Standout extra | Largest ecosystem and API; GPT-6.1 Sol at $2/$10 per million tokens for developers | Claude Code coding agent on paid plans; widely considered the strongest writer | 1,000 video-generation credits + 5 TB storage on AI Pro; YouTube Premium and up to 30 TB on Ultra |
A couple of fine-print notes the table can't hold: OpenAI's Go tier is the only paid plan that may include ads — ad-free starts at Plus. Claude adds sales tax on top of the listed price, and annual billing drops Pro to $17 a month. And on all three, the free tier is genuinely usable for everyday questions; none of them is a demo.
Why it matters
Here's the thing most comparison articles skip: the wrong choice costs you twice. Overbuy and you've prepaid for an agent allowance you'll never touch — ChatGPT Pro at $500 a month buys 25x the Plus usage, which is meaningless if you only write emails. Underbuy and you spend your evenings staring at limit messages mid-project, which is its own tax on your time.
The honest map looks like this:
- ChatGPT is the default pick if you want one subscription that does everything adequately — chat, images, voice, deep research, agents — with the biggest third-party ecosystem and the most generous headline context window.
- Claude is the pick for writers and coders. Its prose quality and Claude Code are the two things power users name when they switch, and the $20 Pro tier is the simplest pricing in the business — no ad-supported middle tier, no five-ladder maze.
- Gemini is the pick if you live in Google's world. Gmail, Drive, Docs, and Photos integration is real, AI Plus at $4.99 is the cheapest meaningful upgrade anywhere, and AI Pro's video credits are a category the other two don't seriously contest.
And the cheapest good answer to "which one" is: start free on all three. They cost nothing to try, and a week of your actual usage tells you more than any comparison table — including this one.
Impact on the USA and the world
These three subscriptions are now some of the most consequential pricing decisions in consumer software. Well over a billion people use ChatGPT weekly; Gemini rides on every Android phone and Google account; Claude has quietly become the developer favorite. The meters decide who gets to think with the best models — students on free tiers, freelancers weighing $20 against groceries, startups choosing an API line item.
The direction of travel favors the buyer: limits have been loosening as models get cheaper to run — that's how OpenAI's everyday chat became unlimited in the first place. The $20 tier that looked premium two years ago is now the middle of the menu, and Google's $4.99 entry is dragging the floor down for everyone.
What happens next
Expect the metered allowances to keep creeping upward while headline prices hold — efficiency gains get passed down as bigger weekly caps, not cheaper plans. Watch for two wildcards: OpenAI's ad-supported Go tier could normalize ads in AI subscriptions (a real fork in the road), and any of the three could follow Google's bundle playbook, folding in storage and video to justify the top tiers.
Our take: don't overbuy. Match the row to what you actually do, not what you aspire to. If you can't name the specific limit you're hitting, you don't need a paid plan yet — and if you can, this table just told you which one. The most expensive tier compared:
Most expensive consumer tier, USD per month. OpenAI's top Pro tier costs 2.5x the other two — aimed at researchers and heavy agent users, not regular subscribers.
Frequently asked questions
Is any of them actually free?
Yes — all three have genuinely useful free tiers for everyday questions. The paid plans buy bigger usage allowances, stronger models, and extras like agents, video generation, or storage. If you only chat casually, free is fine.
Which is best for students?
Start with Gemini's free tier or ChatGPT's free tier — both handle homework-scale work well. If you're hitting limits, Gemini AI Plus ($4.99) is the cheapest upgrade, and Claude Pro is the favorite for long-form writing.
Which is best for coding?
Claude, by the consensus of working developers — Claude Code on the paid plans is the differentiator. ChatGPT's Codex/Work allowance is the alternative if you're already in OpenAI's ecosystem.
Are there annual discounts?
Claude Pro drops to $17/month on annual billing. OpenAI and Google price monthly; watch for promos rather than standing annual rates.
Do the listed prices include tax?
Not always — Anthropic adds sales tax on top of the plan price, for example. Budget a few dollars over the headline number wherever you are.
Related coverage: the full tier-by-tier breakdowns this comparison is built on — ChatGPT Token Limits in 2026: Free vs Plus vs Pro vs API, Claude Free vs Pro vs Max: Message Limits and What You Actually Get, and Gemini Free vs Paid: The Real Limits on Messages, Models, and Video.
Sources: OpenAI — Plans and Pricing (chatgpt.com/pricing) and Models docs (platform.openai.com/docs/models), verified Oct 21, 2026; Anthropic — claude.com/pricing and Help Center, verified Oct 22, 2026; Google — Gemini plan pages, verified Oct 23, 2026; no announced price changes found on Oct 24, 2026. Prices USD, per month, billed monthly; taxes and fees may add to headline prices.