ChatGPT vs Claude vs Gemini: Plans and Prices

The three assistants cost nearly the same where most people buy: about $20 a month (ChatGPT Plus $20, Claude Pro $20 or $17 billed annually, Google AI Pro $19.99). They differ at the edges. Google has the cheapest paid plan at $4.99 and bundles storage and other apps. ChatGPT has the tallest ladder, up to $500. Claude bundles API credits on its $100 and $200 plans. Checked on 31 October 2026.

The short version

Plans side by side

Prices are those on the vendor pages we could read. Taxes are extra on all three. Google's page showed United States prices after we set the region; other regions are priced in local currency.

ChatGPT

OpenAI's Codex pricing page lists the ChatGPT plans that include Codex: Free ($0), Go ($8), Plus ($20), Pro (from $100, with plans at $100, $200 and $500), Business, Edu and Enterprise. Plus includes Codex on the web, CLI, IDE extension and iOS, cloud integrations such as code review and Slack, and the models GPT-6.1 Sol and GPT-6 Luna. Pro adds Ultrafast mode, with access on the $500 plan.

We could not load OpenAI's consumer pricing page on the day we checked, so the non-coding feature list for each ChatGPT plan is not tabulated.

Claude

From claude.com/pricing: Free includes chat on web, desktop and mobile, web search, file creation and code execution, memory, connectors, Artifacts, and Claude Design, Slides and Docs, with limited usage every five hours and every week. Pro ($17 a month annually, $20 monthly) adds Claude Code, Research, Projects, Claude in Chrome, Microsoft 365 and Opus. Max ($100) gives 5x Pro usage and $100 a month in API credits. Ultra ($200) gives 10x Pro weekly usage with no five-hour limit for verified accounts and $200 a month in API credits.

Gemini

From Google's subscriptions page:

Google sells Gemini as part of a bundle. If you already pay for storage or YouTube, the effective price of AI Pro is lower than $19.99; if you do not, those extras are noise.

Models

Prices are per million input and output tokens on each vendor's standard API tier.

Context windows: OpenAI lists 1.05M tokens for its four flagship models. Anthropic lists 1M for all four current models. Google describes Argon as having a unified 1M context window with long decodes up to 1M tokens.

What stands out

The OpenAI and Anthropic ladders match almost price for price, as we show in the OpenAI vs Anthropic API page. Google is the outlier: Argon is listed at a fraction of the other flagships' price, and the page marks that price as a through-November-30 rate that doubles on December 1, 2026. If you are budgeting a project that will run into next year, use the higher price. Google's page also says Argon is not available on the free API tier, and every token counts as billed, including thinking tokens.

Gemini 3.8 Flash has a free API tier (with data used to improve Google's products on the free tier and not on the paid tier), per the pricing page. Grounding with Google Search gives 5,000 free requests a month shared across Gemini 3 and newer models, then $14 per 1,000.

Coding

Each vendor has a coding agent tied to its plans.

See Claude Code vs Codex and Antigravity vs Claude Code for the agent-level comparison.

What each service is best known for

We stay with documented features here, and not with reputation.

ChatGPT. The widest plan ladder, a coding agent on Free, and OpenAI's own model range from Luna to Astra.

Claude. One usage pool across chat and Claude Code, Artifacts and document tools on the Free plan, an annual discount, and API credits bundled into Max and Ultra.

Gemini. Bundles: storage, Gmail, Docs and YouTube tie-ins, image and video generation with Flow credits, and a very low-priced flagship on the API for now.

Data and privacy

Anthropic's data-usage page says Free, Pro, Max and Ultra accounts have a setting that lets data be used to improve future models (five-year retention if allowed), and that commercial plans and the API are not used for training by default.

Google's API pricing page shows a "Used to improve our products" row: Yes on the free tier of a model like Gemini 3.8 Flash, and No on the paid tier. Consumer Gemini app terms are separate. We did not compare the Gemini app's training and retention settings by plan.

OpenAI's rules differ by plan, and we did not compare them here; read its data controls pages for yours.

Whichever you pick, read the account settings page before uploading private work.

How the usage limits work on each

None of the three sells a fixed number of messages, and all three explain why.

Claude says most accounts have a limit that resets on a rolling five-hour session window, with weekly limits on top for paid plans. Chat on web, desktop and mobile and Claude Code share the same pool. How quickly you reach it depends on the model, the length of your chats and the size of your files.

OpenAI's Codex page gives estimates instead of promises: for Plus, a range of local messages per five-hour period for each model, from a few dozen on the heaviest model to thousands on Luna. It says the numbers are not fixed limits, that cloud tasks may use more, and that Pro plans currently have no five-hour limit. When you run out, Plus and Pro users can buy credits.

Google describes its tiers as multiples: AI Plus gives 2x the usage of Free, AI Pro 4x, and AI Ultra up to 20x the Pro limits (the $99.99 option is 5x Pro). The multiples are relative to a Free allowance that Google does not state as a number.

So the three plans are each priced against an allowance you cannot read in advance. The only reliable way to compare is to use them for a week and watch the usage screens.

Switching between them

Your chats do not move, but your habits do. Export what you need before you cancel. Keep prompts, project instructions and reusable files in a folder you own, so a new assistant can start from the same material. For coding, keep an AGENTS.md in the repo: Codex documents it and Claude Code can read it in place of CLAUDE.md.

Which to pick

Choose ChatGPT when:

• You want the most flexible plan sizes, from $8 to $500.
• A coding agent on the free plan matters.
• Your colleagues already use it.

Choose Claude when:

• You want Claude Code in the same subscription as chat.
• You value bundled API credits on higher plans.
• You write long documents and want Artifacts and Docs tools.

Choose Gemini when:

• You already live in Gmail, Docs and Drive and want the integration.
• Storage and YouTube Premium Lite are worth something to you.
• You want the cheapest paid entry at $4.99.

Use more than one when:

• You do high-stakes work and want a second opinion. The $20 tiers are a cheap way to run the same prompt on all three.

How to run a three-way test

1. Write three tasks from your actual week: one document, one analysis, one coding or spreadsheet task.
2. Use the same prompt and the same files on the standard tier of each.
3. Hide the labels and ask someone else to rank the results.
4. Count the corrections needed.
5. Check each usage page to see how much allowance the test consumed.

If you cannot decide after that, take the one with the plan features you will use weekly, not the one with the longest feature list.

A short checklist

• Free plan first: all three have one.
• Decide your price tier: $5, $8, $20, $100 or above.
• Do you need a coding agent, storage or API credits?
• Is the training setting off or on in your account?
• Are you budgeting past December 1 for Gemini 4 Argon?

What a year costs

Monthly prices hide the real commitment. This table multiplies the list prices above by twelve, with Claude Pro also shown at its annual billing.

At the standard tier the three land within a few dollars of each other. Only Claude's annual billing saves anything, $40 over monthly billing. At the high and top tiers the price points line up closely again, and ChatGPT adds a $500 option that the others do not list. Taxes are extra, and we did not check each vendor's refund and cancellation terms.

So a year of standard-tier use costs about $240 whichever you choose. That is the budget for the question "which one", and it is small next to the time you spend picking tools. Treat the first month as the real trial and cancel the two you do not use.

Matching a plan to a person

A student or someone trying AI for the first time. All three have a free plan. Start there and note which limit you hit first.

A writer or analyst working with long documents. The $20 tier on any of the three, with a three-way test on your own files. Claude's Free and paid plans list file creation, Artifacts and Docs tools, and Gemini lists Deep Research and Gemini Notebook, so compare those against what you actually do.

A developer. Claude's plans include Claude Code, and OpenAI's plans include Codex. Gemini AI Pro lists entry-level rate limits for the agent model in Antigravity. See Claude Code vs Codex and Antigravity vs Claude Code.

Someone already inside Google's apps. Gemini's storage and Gmail and Docs integration are bundled, so the effective price is lower if you would pay for storage anyway.

A small team. Check the Team and Business plans on each vendor's page before buying individual seats. The Claude Team plan lists $20 and $100 seats in the table above.

Questions people ask

Which is cheapest? Gemini's AI Plus at $4.99 is the cheapest paid entry in the table. ChatGPT Go at $8 is next. Claude has no plan between free and Pro.

Can I run all three? Yes. Three standard plans cost about $720 a year at list price, which is why we suggest a one-month trial and then dropping two.

Do the plans share usage with coding tools? Claude says chat and Claude Code share one pool. OpenAI says Codex shares usage with ChatGPT. Check Google's page for how Antigravity usage is counted against AI Pro.

Our take

Going by the documents, price is not what separates these three at the standard tier: $20 buys roughly the same thing everywhere, and the API ladders for OpenAI and Anthropic match to the dollar. What differs is what comes bundled. Google bundles storage and apps. OpenAI puts a coding agent on the free plan. Anthropic puts one usage pool under chat and Claude Code, and adds API credits at the top.

So choose by the bundle you will use, then by a trial on your own work. We did not run model tests for this page, and we have marked the ChatGPT and Gemini consumer details we could not read. Everything else here is read off the vendors' own pricing and model pages on 31 October 2026, and prices on all three change often.

Related

• ChatGPT vs Claude
• Claude Opus vs Sonnet
• Claude API vs OpenAI API
• Compare AI models
• Antigravity vs Claude Code
• How we compare