Claude vs ChatGPT: which plan is cheaper?

Claude vs ChatGPT subscription cost, side by side

The Claude vs ChatGPT subscription cost question has a short answer. Across the middle of the range the two price ladders are almost the same, so price alone will not decide it for you. Claude Pro and ChatGPT Plus both cost $20 a month. Each company also sells a heavy tier at $100 and a heavier one at $200. The ladders split at the top: OpenAI also sells a ChatGPT Pro plan at $500, and Anthropic's individual plans stop at $200. The other gap sits at the bottom of the ladder, where OpenAI sells an $8 tier called Go and Anthropic sells nothing between Free and Pro.

Every figure on this page is a US list price. The Claude prices and the ChatGPT Go, Plus, Pro and Business prices were rechecked on 31 October 2026. Both companies change these prices. Read the vendor page before you pay.

[
  {
    "plan": "Claude Free",
    "usd_price": 0
  },
  {
    "plan": "ChatGPT Free",
    "usd_price": 0
  },
  {
    "plan": "ChatGPT Go",
    "usd_price": 8
  },
  {
    "plan": "Claude Pro",
    "usd_price": 20
  },
  {
    "plan": "ChatGPT Plus",
    "usd_price": 20
  },
  {
    "plan": "Claude Max",
    "usd_price": 100
  },
  {
    "plan": "ChatGPT Pro $100",
    "usd_price": 100
  },
  {
    "plan": "Claude Ultra",
    "usd_price": 200
  },
  {
    "plan": "ChatGPT Pro $200",
    "usd_price": 200
  },
  {
    "plan": "ChatGPT Pro $500",
    "usd_price": 500
  }
]

That is 10 tiers, and from $20 up to $200 the two ladders match each other step for step. Both companies are pricing against the same guess about what one person will pay for one seat, so they land on the same numbers. The chart stops at individual seats, so if you want the Anthropic side of the ladder in full, every Claude plan with its price and billing period puts Team, Enterprise and the API in the same table and marks which ones include Claude Code.

Why a messages per month number does not tell the whole story

You cannot work out a reliable cost per message for a whole subscription from a single allowance. Anthropic's help pages say usage depends on the length and complexity of your conversations, the features you use, which model you are talking to, and the effort level you picked. Its Pro page says the number of messages varies with message length, the files you attach, the length of the conversation and the model you use. Anthropic publishes no fixed message count, but OpenAI publishes model-specific Chat allowances: Business Standard includes 15 Pro messages per month, shared across GPT-6 Pro and GPT-5.6 Sol Pro, while Business Premium includes 50 Pro messages per week shared across those models. These Chat allowances are separate from variable Work and Codex usage. For Plus and Standard Business in Codex and ChatGPT Work, its help centre gives estimated ranges of local messages per five hour window, from 5 to 45 on GPT-6 Astra up to 350 to 3,000 on GPT-6 Luna, and adds that these ranges are estimates, not guaranteed message counts. A range that wide is still not a number you can divide a price by.

Claude and OpenAI's Work and Codex measure usage over usage windows rather than a single monthly message quota. Anthropic describes Max as five times Pro usage every five hours and every week. Ultra gives ten times Pro usage every week, with no five hour limit for verified accounts; accounts not yet verified have a temporary limit. Ultra has a weekly limit across all models and a separate weekly Fable limit. On the OpenAI side the five hour window belongs to Codex and ChatGPT Work. Those Plus estimates apply there, and OpenAI's Codex pricing docs say Pro plans currently have no five hour limit, with weekly limits that may also apply. Chat is counted separately. OpenAI states that usage and credit rules in Work and Codex are separate from Chat, and Chat runs on model-specific allowances, including monthly, weekly and daily limits on its Pro models. This is why two people paying the same $20 can have completely different experiences of the same plan.

So compare the two on your own workload. How Claude counts a usage limit explains the mechanism on the Anthropic side, and what to do when you hit that limit covers the part that actually costs you time.

What the free plans give you

Claude Free costs $0. As of 25 October 2026 Anthropic's pricing page lists chat on web, desktop and mobile, plus web search, file creation, running code, memory across conversations, connections to your apps and tools, Artifacts, and Claude Design, Slides and Docs. Claude Code, the coding tool that runs in your terminal, is not part of the free tier. That single exclusion is what moves most developers onto a paid seat.

ChatGPT Free costs the same, but it no longer has the same shape. On 6 August 2026 OpenAI announced unlimited everyday text chats for Free and Go users, switched on from the following week and subject to abuse-prevention safeguards. As of 25 October 2026 OpenAI's help centre still lists them for both plans. Uploads, image generation, voice and other tools keep their own separate limits. Codex, OpenAI's coding tool, is available on the free tier, where OpenAI's Codex pricing page lists GPT-6 Luna at standard speed in the desktop app, subject to rollout.

Free is the correct answer for a lot of people. If you ask a handful of questions a day and you do not write code, stay on free and spend the money somewhere else. What the Claude free tier actually allows is worth reading before you assume you need to pay.

Where ChatGPT Go fits, and why Claude has no answer to it

ChatGPT Go costs $8 a month. It sits between Free and Plus on price. Anthropic sells nothing at that price. If your budget is genuinely under ten dollars a month, ChatGPT is cheaper, because it is the only one of the two selling at that level. OpenAI's August 2026 change gave Free and Go users the same unlimited everyday text chats, so if plain text chat is all you do, check what Go adds that you would actually use before you pay for it. Whether ChatGPT Go is worth it next to Free and Plus walks through what Go adds over Free and what you still give up against Plus, such as ad-free chats and the reasoning models.

This is the clearest price difference in the whole comparison, and it stops mattering the moment your budget reaches $20. From there up to $200 the two ladders sit on top of each other. If you pay in a currency where the US price converts badly, how Claude Code pricing lands outside the US is the next thing to check. Settle the currency before you commit, because switching Claude to a different billing currency means canceling at the end of your term and resubscribing, so an annual seat keeps the currency you started with for the whole year.

Claude Pro against ChatGPT Plus at the same price

Both cost $20 a month, so the decision is about what comes in the box. Two differences carry real money.

Claude Pro includes Claude Code. Anthropic states that Pro, Max and Ultra cover the Claude apps and Claude Code in the terminal under one subscription, and that the usage limits are shared across both. Shared is the important word: a long coding session draws on the same allowance your chat window uses, so a heavy afternoon in the terminal can close the chat window too.

Claude Pro can be billed for a year up front. ChatGPT Plus cannot. As of 25 October 2026 Anthropic sells a Pro seat for twelve months at a discount, while OpenAI's help centre says it does not support annual billing for Plus and that Pro is billed monthly too. A Plus seat is a monthly charge, every month.

[
  {
    "plan": "Claude Pro, billed annually",
    "usd_annual_cost": 200
  },
  {
    "plan": "Claude Pro, billed monthly",
    "usd_annual_cost": 240
  },
  {
    "plan": "ChatGPT Plus, billed monthly",
    "usd_annual_cost": 240
  }
]

Paying yearly moves a Claude Pro seat from $240 to $200 over twelve months. That is a real saving and a real commitment, so take it after a month of paying monthly, once you know you still want the seat. Before you lock in the year, how annual billing behaves once you are on it is worth a look, because it sets out how an upgrade part way through the year is prorated and what you get back if you cancel early. Everything else at this price point is a judgement about the models and the interface, and you can test that for nothing on both free tiers. The Claude Pro price and where its limits sit goes further on the Anthropic side. If you have already settled on Claude and the only question left is whether to pay at all, deciding whether Claude Pro is worth it starts from how often the free limit stops you rather than from the feature list.

When is a Max or ChatGPT Pro tier cheaper than staying put?

Above the base tier both companies sell bigger allowances. Claude Max is $100 for five times Pro usage every five hours and every week, while Ultra is $200 for ten times Pro usage every week. Ultra replaces Max 20x, and existing subscribers move automatically with no price change. Max and Ultra are separate plans, both billed monthly only, with $100 and $200 a month in API credits respectively, claimed into a linked Claude Console organization. ChatGPT Pro is sold at three prices: $100, $200 or $500 a month, and only the $500 plan includes Ultrafast mode. OpenAI's help centre ranks the three by included usage, with Pro 500 the highest, but states no multiple of Plus for any of them. At its 29 September 2026 launch, OpenAI put Pro 500 at 25 times the ChatGPT Plus allowance; it has since doubled Pro 100 and Pro 500 Work and Codex allowances. The planned Pro 200 reduction was canceled: existing subscribers with the higher allowance keep it, and lower allowances have doubled to match, including for new subscriptions, at the same price. Read OpenAI's Pro plans page before you pick one.

Read any multiplier carefully, because each one is measured against its own vendor's base plan. Five times Pro and five times Plus would not be the same quantity of work, since Pro and Plus do not start from the same allowance and neither Claude Pro usage nor Plus Work and Codex usage is published as a fixed message count. Treat a multiplier as a direction of travel, not as a measurement you can compare across companies. The $500 ChatGPT Pro plan has no Claude counterpart, because Anthropic's individual plans stop at Ultra. If you are weighing the two $200 tiers against each other, Ultra set against ChatGPT Pro compares what that money buys on each side, whether the coding agent comes with the seat, and how each one behaves once you hit the ceiling.

The test that works is simple. Count the hours you lost last month waiting for a limit to reset. If those hours are worth more to you than the price of the upgrade, upgrade. If you have never been blocked, you do not need the tier, whatever the plan page says about it. Trying a heavier tier for a month is a small risk, because canceling Max or Ultra keeps access through the billing period you have already paid for. Max against Ultra works through that decision for Claude, and which Claude plan you actually need covers the full ladder.

Is paying for Claude worth it for coding?

This is the question people keep asking around this comparison, so here is the honest version.

On the Anthropic side, Claude Code arrives with Pro, Max and Ultra at no extra charge, and its usage comes out of the same pool as the chat apps. One payment, one pool, one number to watch. That is easy to reason about, and it is the main reason developers pick the Claude subscription over the ChatGPT one.

On the OpenAI side, Codex is available across the ChatGPT plans, including Free and Go, with the allowance rising as the tier rises. OpenAI now counts Codex usage in credits priced per million input, cached input and output tokens rather than in messages. Plus and Pro subscribers who run out can buy more credits rather than move up a tier. Claude Pro, Max and Ultra subscribers can also enable usage credits to continue beyond their included limits at standard API rates. On either product, a heavy week can cost more than the subscription itself.

The practical advice is the same for both products. Start on the base tier of whichever one you prefer using. Work for a month. Upgrade only if being blocked breaks your concentration often enough that you notice. What Claude Code costs on a subscription has the detail, how Claude Code spends tokens explains what drives the number up, and how Claude Code compares with Codex and Copilot covers the tools rather than the bill.

Model choice matters more here than plan choice. Both products give you a small fast model and a large slow one, and sending routine work to the small one changes your usage more than an upgrade does. Which Claude model to run for which job is that decision for Claude.

Buying for a team rather than for yourself

Team pricing matches across the two products. As of 31 October 2026 a Claude Team standard seat is $25 a month billed monthly, or $20 billed annually. ChatGPT Business matches both of those numbers. Anthropic then adds a premium seat at $125 monthly or $100 annually, carrying five times the usage of a standard seat. ChatGPT Business now also offers Premium seats at those prices, with higher Work and Codex allowances and no five hour usage limit.

[
  {
    "plan": "Claude Team standard, annual",
    "usd_seat_price": 20
  },
  {
    "plan": "ChatGPT Business, annual",
    "usd_seat_price": 20
  },
  {
    "plan": "Claude Team standard, monthly",
    "usd_seat_price": 25
  },
  {
    "plan": "ChatGPT Business, monthly",
    "usd_seat_price": 25
  },
  {
    "plan": "Claude Team premium, annual",
    "usd_seat_price": 100
  },
  {
    "plan": "Claude Team premium, monthly",
    "usd_seat_price": 125
  },
  {
    "plan": "ChatGPT Business Premium, annual",
    "usd_seat_price": 100
  },
  {
    "plan": "ChatGPT Business Premium, monthly",
    "usd_seat_price": 125
  }
]

A team seat buys central billing, single sign-on and admin controls. Those are the reason to move a small company off individual seats, and they matter more than the handful of dollars between the two vendors. Claude Team plans for a small business covers the seat maths. The step above Team stops behaving like a seat price at all, so read what a Claude Enterprise seat actually bills you for before you assume Enterprise is just Team with more people on it. Once you are weighing that step for a whole company rather than a handful of people, Claude and ChatGPT compared at the enterprise tier is the more useful comparison, because only one of the two publishes a seat price at that level and the rest of the decision turns on seat minimums, SCIM, audit logs and retention terms.

Where this comparison stops

This page compares monthly seats. Two neighbouring questions have their own answers, and mixing them up is how people end up paying twice.

If you want cost per million tokens, that is a different product with a different bill. Claude and ChatGPT API pricing per million tokens has the rates, and those rates move independently of the subscription prices here. Rates also vary sharply from one model to the next, so if a single model is doing most of your work, what Claude Fable 5 costs per million tokens and the jobs it suits shows how to read one model's published rate against the work you send it.

If you are asking whether to skip the subscription and pay for API calls instead, the API against a subscription works through the break-even point. The short version: a subscription is a fixed cost with a usage ceiling, and API billing is a variable cost subject to spend and rate limits. Before you put a card on an API account to find out which way your own work falls, check whether the Claude API has a free tier. Max and Ultra subscribers can claim their included monthly API credits without adding a payment card to the Claude Platform.

How to choose in ten minutes

• Use both free tiers for a week on your real work, not on test questions you already know the answer to.
• If you write code most days, start with Claude Pro, because Claude Code is included in the seat.
• If your budget is under ten dollars a month, take ChatGPT Go, since Anthropic sells nothing at that level, but try ChatGPT Free first if plain text chat is most of what you do.
• Pay monthly first. Move to annual billing after a month you were glad to have paid for.
• Move up to a $100 tier only after you have been blocked by limits often enough to count the hours.

FAQ

Is Claude cheaper than ChatGPT?

At most tiers they cost the same. Claude Pro and ChatGPT Plus are both $20 a month, and both companies sell heavier tiers at $100 and $200, checked on 25 October 2026. Above that, OpenAI sells a ChatGPT Pro plan at $500 a month with Ultrafast mode, while Anthropic's individual plans stop at $200. ChatGPT is cheaper at the bottom: it sells a Go tier at $8 a month, and Anthropic has nothing between Free and Pro. Claude is cheaper in one place: Pro can be billed annually at $200 for the year, while an individual ChatGPT seat is sold monthly.

Is paying for Claude worth it if I mostly write code?

Claude Code is included with Pro, Max and Ultra, and its usage comes out of the same allowance as the Claude apps, so one subscription covers the terminal and the chat window. On the ChatGPT side Codex is included across the plans, and its usage is now counted in credits priced per million tokens rather than in messages, so Plus and Pro subscribers can buy more credits instead of upgrading. Start on the base tier of whichever you prefer, work for a month, and upgrade only if hitting the limit is costing you real time.

Do I need a Max or ChatGPT Pro tier?

Only if the base tier is stopping you. Claude Max gives five times Pro usage every five hours and every week at $100, while Ultra gives ten times Pro usage every week at $200. ChatGPT Pro is sold at $100, $200 and $500. OpenAI's help centre ranks them by included usage without stating a multiple of Plus; its launch announcement put Pro 500 at 25 times the Plus allowance, before its Work and Codex allowance doubled. Because any multiplier is measured against the vendor's own base plan, five times Pro and five times Plus would not be the same quantity. Count the hours you actually lost last month waiting for a limit to reset, and buy the upgrade only if those hours are worth more than it costs.

Can I save money by paying for a year up front?

On Claude, yes. As of 25 October 2026 Anthropic sells a Pro seat at $200 for twelve months, against $240 if you pay month by month. OpenAI's help centre says Plus has no annual billing and Pro is billed monthly, so there is no yearly discount on an individual ChatGPT seat. Pay monthly for the first month either way, because an annual plan you stop opening in week three is the most expensive option on this page.