ChatGPT vs Claude: Plans, Models, Prices

Both assistants cost the same at the entry level: $20 a month for ChatGPT Plus and for Claude Pro. ChatGPT has a cheaper $8 Go plan and a wider range of top tiers up to $500, and its Free plan includes a coding agent. Claude's plans stop at $200 and Claude Code starts on Pro. The better assistant for you depends on the work, which this page leaves to your own trial. Plans and prices were checked on 31 October 2026.

The short version

Plans side by side

ChatGPT

OpenAI's Codex pricing page lists the ChatGPT plans, because Codex is part of them. The headline prices, from that page:

The page says Codex and ChatGPT Work are included in the Free, Go, Plus, Pro, Business, Edu and Enterprise plans and share usage. Plus adds Codex on the web, in the CLI, in the IDE extension and on iOS, cloud integrations such as automatic 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 were not able to read OpenAI's consumer ChatGPT pricing page from our tools on the day we checked, so consumer features beyond Codex (voice, image generation limits, memory, file tools) are not tabulated here.

Claude

From claude.com/pricing:

Free includes chat on web, desktop and mobile, web search, file creation and code execution, memory across conversations, connected apps, Artifacts, and Claude Design, Slides and Docs. Claude Code, Research, Claude in Chrome and Microsoft 365 integration start on Pro.

What the price ladders tell you

Both vendors have a $20 plan, a $100 plan and a $200 plan. ChatGPT adds an $8 plan below and a $500 plan above. Claude's cheaper route is the annual discount on Pro, which brings it to $17 a month.

The meters are similar in style. Claude uses a rolling five-hour window plus weekly limits, with a single pool across chat and Claude Code. OpenAI describes estimates of local messages per five-hour period for Codex on Plus, with weekly limits that may also apply, and no five-hour limit on Pro. Neither publishes a fixed number of chats, because usage depends on the model, the length of the conversation and the tools used.

Models

Anthropic

Anthropic's models overview lists four current models:

All four list a 1M-token context window, 128K max output and a June 2026 knowledge cutoff. On the consumer plans, Free includes Sonnet and Haiku, Pro adds Opus, and Fable is paid with usage credits on Pro and included up to half your weekly limit on Max and Ultra. See Opus vs Sonnet for the detail.

OpenAI

OpenAI's models page lists these flagship models:

Each lists a 1.05M-token context window and 128K max output. Knowledge cutoffs run from April to June 2026, depending on the model.

Reading the two tables together

The tiers line up almost one to one, down to the price. Astra and Fable are both $10 and $50. Sol and Sonnet are both $2 and $10. Luna and Haiku are both about $0.10 and $0.50 at the low end. Opus 5.5 at $4 and $20 sits in the middle, and Terra at $1 and $5 sits between Sol and Luna.

That makes price a weak reason to pick one over the other at the API level. The interesting differences are behavior, tools and ecosystem, and those need testing on your work. We do not rank the models on quality: both vendors publish benchmark charts run on their own settings, and we did not run any ourselves.

Features that differ

We have left the ChatGPT cells that we could not read marked, instead of guessing.

Coding

This is where the pages above give the most concrete comparison, since both vendors document an agent.

Codex is a CLI under Apache-2.0, an IDE extension for VS Code, Cursor and Windsurf, a desktop app and a web agent. It signs in with a ChatGPT account, so a Plus subscription covers it. A codex exec command runs it in scripts and CI.

Claude Code runs in the terminal, in VS Code and JetBrains, in a desktop app and on the web, backed by one engine. It has skills, hooks, subagents and plugins, plus claude -p for scripting. It is included from Pro.

Our full comparison is in Claude Code vs Codex.

Data and privacy

Anthropic's data-usage page is specific. For Free, Pro, Max and Ultra accounts there is a setting that lets your data be used to improve future models, and the page says that includes Claude Code use. Users who allow it are subject to a five-year retention period. For Team and Enterprise plans, the API and third-party platforms, Anthropic says it does not train generative models on code or prompts sent to Claude Code under commercial terms, unless the customer opts into a program.

For ChatGPT, the controls vary by plan and have their own pages.

In both cases the practical step is the same: open the settings page on the account you pay for, and decide whether the training option is on before uploading anything private.

Teams

Claude's Team plan is $20 per standard seat and $100 per Premium seat a month, billed annually, for 2 to 150 people, and includes Claude Code on every seat with pooled API credits up to $500 a month. Enterprise adds SCIM, audit logs, custom data retention and role-based access, at $20 per seat plus usage at API rates.

ChatGPT has Business, Edu and Enterprise. On Codex, the pricing page says standard Business has the same per-seat estimates as Plus, and Enterprise and Edu plans without flexible pricing have the same per-seat usage limits as Plus for most features, while flexible-pricing Enterprise and Edu accounts have no fixed rate limits and usage scales with credits.

If your team needs a security review, ask both vendors for the same questionnaire. The public pages do not settle it.

Who each suits

Pick ChatGPT when:

• The $8 Go plan fits your budget better than $17 or $20.
• You want a coding agent on the free tier.
• You want a high-usage individual tier above $200.
• Your company already has a ChatGPT workspace.

Pick Claude when:

• You want Claude Code in the same plan as chat, and you will use it daily.
• You want monthly API credits bundled with a subscription (Max, Ultra).
• Documents, Slides and Design on the Free plan appeal to you.
• You prefer the Claude model family's writing and agent behavior. That is a preference we cannot test for you.

Pay for both when:

• You want a second opinion on hard questions and the cost of a second $20 plan is small against the value of your time.

A fair way to test them

Neither vendor's benchmark page will tell you which suits your work. Do this instead.

1. Pick five real tasks: two writing, two analysis, one coding.
2. Run each on the same-tier plan on both services, using the same prompt.
3. Score the outputs blind, if you can, or ask a colleague to.
4. Note how many times you had to correct each one.
5. Check your usage page afterward to see how much of the allowance each consumed.

That takes an afternoon and costs one month at $20, or nothing on the Free plans.

A short checklist

• Which price tier do you need: $8, $20, $100 or above?
• Do you need a coding agent, and in which environment?
• Is the training setting on or off in your account?
• Do you need an API too? Compare the tier table, not the logos.
• Will a team of five or fifty use it? Compare Team and Business plans.

Individual cost over a year

Multiplying the list prices above by twelve shows what each step up actually commits you to.

Claude's annual plan is the only discount in either column, $40 a year on Pro. The jump from standard to high is $100 a month on both sides, and what you buy for it differs: Claude Max lists 5x Pro usage and $100 a month in API credits. ChatGPT's Pro page lists plans at $100, $200 and $500, with no five-hour limit on current Pro plans for Codex. Before you step up, read your usage screen. If you are not hitting the standard tier's limit at least weekly, the next tier buys you headroom you will not use.

Team seat math

Claude's Team plan lists a Standard seat at $20 and a Premium seat at $100 per seat a month billed annually, for 2 to 150 people. Enterprise is $20 per seat a month plus usage at API rates.

ChatGPT Business is priced per seat on OpenAI's page, but we could not read a figure, so we do not total it. The useful exercise for a Claude team is to start everyone on Standard and promote only the people who hit the limit, since a Premium seat is five times the price.

A working week with each

A fair test is a normal week. On Monday, give both the same long document to summarise and critique. On Tuesday, ask each to turn a messy spreadsheet or CSV into a clear table with a short written finding. On Wednesday, give each a small coding job, using Codex for ChatGPT and Claude Code for Claude if both are on your plan. On Thursday, use each for a task where you need a source, and check every source it names. On Friday, look at the usage screen of each and write down how much of the allowance the week used.

Score each day on three things only: whether you had to correct it, whether you trusted the sources, and whether you would hand the output to someone else. That is a small experiment, and it will tell you more than any feature list.

Questions people ask

Is one smarter? We do not rank them. Both vendors' API tiers match in price, and quality on your task needs your own test.

Can I move my chats? Not directly. Keep your prompts and project instructions in files you own.

Which has a free coding agent? ChatGPT Free includes Codex for quick tasks. Claude Free does not include Claude Code.

Our take

Going by the documents, the two services are closer than their fans suggest. They price their chat plans the same at the middle ($20, $100, $200), they price their API tiers almost identically, and both ship a coding agent included in the subscription. The real differences are at the edges: ChatGPT has the cheapest paid plan ($8) and the tallest ladder ($500) and puts a coding agent on Free; Claude has the annual discount ($17), bundled API credits on Max and Ultra, and a stronger documented extension system around Claude Code.

If money is the only question, start with ChatGPT Go or the Free plans of either. If a coding agent in your terminal is part of the question, compare Claude Code and Codex directly, since that choice affects your work more than the chat window does.

We did not test either assistant for this page, and several ChatGPT consumer details are marked above because we could not read them. Use the trial plan, and spend an afternoon on the five-task test.

Related

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