# Claude Code pricing explained, with what a real task costs Authors: AutoRouter Author profile: https://autorouter.com/about Published: 2026-11-10T13:55:00.000Z Reading time: 12 min Canonical: https://autorouter.com/blog/claude-code-pricing Part of: How to cut Claude Code and Codex costs (https://autorouter.com/blog/cut-claude-code-and-codex-costs.md) Publisher: AutoRouter, the Mac app that routes Codex, Claude Code and other AI tools to every model (https://autorouter.com) About AutoRouter: a Mac app (macOS 14 or later) that adds every leading model to Codex, Claude Code, Claude Desktop, T3 and Terminal and picks the model for each task, using your ChatGPT and Claude plans first. Install from https://autorouter.com/AutoRouter.dmg, or ask your agent to follow https://autorouter.com/setup.md. Routing across your plans is free; Pro is $50 a month with $20 of included usage. Unrelated to OpenRouter's Auto Router. > As of November 9, 2026, Claude Code comes with Claude Pro at $20 a month ($17 billed annually), Max at $100 and Ultra at $200, or bills per token through an API key, where Claude Opus 5.5, its default model, costs $4 per million input tokens and $20 per million output tokens. No plan publishes a token allowance: every request counts against limits shared with claude.ai. In tests by AutoRouter, the Mac app for Codex and Claude Code, a real bug fix on Opus 5.5 cost $0.15 to $0.22 at API-key prices. ## Key facts - As of November 9, 2026, Claude Code is included in Claude Pro ($20 a month, or $17 billed annually), Max ($100) and Ultra ($200), and isn't part of the Free plan ([Anthropic](https://claude.com/pricing)). - Max 20x is now called Ultra: $200 a month for 10 times Pro's weekly usage, with no five-hour limit for verified accounts ([Anthropic](https://support.claude.com/en/articles/17728952)). - Claude Code's default model on every plan and on the API is Claude Opus 5.5, at $4 per million input tokens and $20 per million output tokens ([model config](https://code.claude.com/docs/en/model-config), [pricing](https://platform.claude.com/docs/en/about-claude/pricing)). - In AutoRouter's tests on November 9, 2026, four real bug fixes in Claude Code on Opus 5.5 averaged $0.184 each at API-key prices, and $0.225 as Claude Code priced them on a Max plan ([method](/studies/codex-vs-claude-code/method.md)). - Outside tests in October 2026 put a $200 Claude plan at about $12,000 a month of Opus 5.5 usage at API prices ([SemiAnalysis](https://semianalysis.com/weekly/episodes/ep-036/)) and Pro at about $673 a month ([Wokey](https://wokey.ai/claude-usage-limits)); SemiAnalysis also estimates Anthropic books 50% to 60% gross margins on its plans. As of November 9, 2026, Claude Code comes with a Claude plan, [Pro at $20 a month](https://claude.com/pricing) ($17 a month billed annually), Max at $100 or Ultra at $200, or bills per token through a Claude Console account and API key. On the API, Claude Opus 5.5, Claude Code's [default model](https://code.claude.com/docs/en/model-config) on every plan, costs [$4 per million input tokens and $20 per million output tokens](https://platform.claude.com/docs/en/about-claude/pricing). Plans publish no token allowance: each request counts against a five-hour limit and a weekly limit shared with claude.ai. In tests on real bug reports from an open-source project, a fix on Opus 5.5 cost $0.15 to $0.22 at API-key prices. [AutoRouter](https://autorouter.com), the Mac app for Codex and Claude Code, ran those tests and wrote this post. ## How much does Claude Code cost on each plan? Claude Code is included in every paid Claude plan, from Pro at $20 a month to Ultra at $200, and isn't part of the Free plan, as of November 9, 2026 ([Anthropic](https://claude.com/pricing)). Claude Code works the same on every plan; the plans differ in how much usage they hold. Max and Ultra are billed monthly only. | Plan | Price (November 9, 2026) | Usage, as Anthropic describes it | | ------------------------ | ------------------------------------------------------------------------------------------------------------ | --------------------------------------------------------------------------------------------------- | | Free | $0 | Claude Code not included | | Pro | [$20 a month, or $17 a month billed annually](https://claude.com/pricing) | Limits every five hours and every week; Claude Fable only through usage credits | | Max | [$100 a month](https://support.claude.com/en/articles/11049741-what-is-the-max-plan) | 5× Pro's usage every five hours and every week; Fable up to half the weekly limit | | Ultra (formerly Max 20x) | [$200 a month](https://support.claude.com/en/articles/17728952) | 10× Pro's weekly usage, no five-hour limit for verified accounts; Fable up to half the weekly limit | | Team, Standard seat | [$20 per seat a month billed annually, $25 monthly](https://claude.com/pricing) | More usage than Pro | | Team, Premium seat | [$100 per seat a month billed annually, $125 monthly](https://claude.com/pricing) | 5× a Standard seat | | Enterprise | [$20 per seat a month billed annually, plus usage at API rates](https://claude.com/pricing), 20-seat minimum | Usage billed per token; admins set spend limits | | Claude Console (API key) | [Per token, at API rates](https://platform.claude.com/docs/en/about-claude/pricing) | Every token is billed instead of counting against plan limits | Each tier now buys the same weekly usage per dollar: Max costs 5 times Pro for 5 times the usage, and Ultra costs 10 times Pro for 10 times the weekly usage. Ultra replaced Max 20x, and [existing Max 20x subscribers moved to Ultra](https://support.claude.com/en/articles/17728952) with no price change. Max and Ultra also include monthly API credits for the Claude Platform, $100 and $200, but those [don't apply to Claude Code](https://support.claude.com/en/articles/12429409-manage-usage-credits-for-paid-claude-plans). For how Claude Code compares with OpenAI's agent on price, see Codex pricing explained, with what a real task costs. ## What do Claude Code's usage limits mean in practice? Anthropic publishes no message or token counts for Claude Code's limits: it says how fast you reach one [depends on the model, the length of your chats and the size of your files](https://claude.com/pricing). Claude, Claude Desktop, the mobile apps and Claude Code all draw from the same pool, so a long chat on claude.ai leaves less for coding. - **Pro and Max** have a five-hour session limit and a weekly limit ([Anthropic](https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan)). - **Ultra** has a weekly limit and no five-hour limit once the account is verified; an unverified account has a temporary limit ([Anthropic](https://support.claude.com/en/articles/17728952)). - **Claude Fable** runs on usage credits on Pro, and on Max and Ultra it has its own weekly limit, up to half the overall one ([Anthropic](https://claude.com/pricing)). - **At a limit,** you can wait for the reset, upgrade, or turn on [usage credits](https://support.claude.com/en/articles/12429409-manage-usage-credits-for-paid-claude-plans), which bill at standard API rates up to a monthly spend limit you set. To see where you stand, run `/usage` in Claude Code. On a plan it shows usage bars and a breakdown of what counts against your limits; its dollar figure is a list-price estimate that [isn't relevant for billing](https://code.claude.com/docs/en/costs) on Pro, Max or Ultra. [How Claude Code usage limits work, and how to make them last](/blog/claude-code-usage-limits) covers the five-hour and weekly windows in detail, and [How to check your Claude Code usage](/blog/check-claude-code-usage) keeps the numbers on screen while you work. ## What does Claude Code cost on the API? With an API key, Claude Code bills every token at Anthropic's list prices, which as of November 9, 2026 run from $0.10 per million input tokens on Claude Haiku 5.5 to $8 on Claude Fable 5.2, released that day ([pricing](https://platform.claude.com/docs/en/about-claude/pricing)). Claude Code sends the conversation with each request and caches it, so most input is billed as cache reads and cache writes, which have prices of their own. | Model, per million tokens | Input | 5-minute cache write | 1-hour cache write | Cache read | Output | | ----------------------------------------------- | ----- | -------------------- | ------------------ | ---------- | ------ | | Claude Fable 5.2 | $8 | $10 | $16 | $0.20 | $40 | | Claude Opus 5.5 | $4 | $5 | $8 | $0.20 | $20 | | Claude Sonnet 5.5 | $2 | $2.50 | $4 | $0.10 | $10 | | Claude Haiku 5.5 (prompts up to 100,000 tokens) | $0.10 | $0.125 | $0.20 | $0.01 | $0.50 | Which cache write you pay depends on how you sign in. With an API key, Claude Code [uses the five-minute cache](https://code.claude.com/docs/en/prompt-caching#which-ttl-each-request-gets), whose writes cost 1.25 times the input price; on a subscription it requests the one-hour cache, at twice the input price. Fast mode for Opus 5.5 costs [twice the standard price](https://claude.com/pricing), and US-only inference adds [10%](https://platform.claude.com/docs/en/about-claude/pricing). Unless you change the model, every request runs on Opus 5.5. ## What does a real Claude Code task cost? A Claude Code task on Opus 5.5 cost between $0.056 and $0.306 at API-key prices in AutoRouter's two published studies, depending on how much code it read and wrote. Each figure is the mean of three fresh `claude -p` sessions with no router in between. The first column is the cost Claude Code reported on a Max plan, which prices cache writes at the one-hour rate. The second reprices the same tokens with the five-minute cache writes an API key uses. | Task on Claude Opus 5.5 | Repository | As Claude Code priced it on Max | With an API key | | ------------------------------------------------------- | ------------ | ------------------------------- | --------------- | | Write a commit message | itsdangerous | $0.082 | $0.061 | | Fix a planted failing test | itsdangerous | $0.078 | $0.056 | | Rename a function across 6 files | itsdangerous | $0.097 | $0.072 | | Write tests for one module | itsdangerous | $0.125 | $0.098 | | Refactor helpers into a new module | itsdangerous | $0.168 | $0.132 | | Fix a real bug: a parameter named `help` | click | $0.237 | $0.196 | | Fix a real bug: `version_option` ignores `package_name` | click | $0.264 | $0.221 | | Fix a real bug: environment variable hints | click | $0.212 | $0.169 | | Fix a real bug: an empty usage line | click | $0.185 | $0.150 | | Write tests for `_textwrap.py` | click | $0.363 | $0.306 | The five itsdangerous tasks ran on October 28, 2026 in a 1,176-line library ([method](/studies/claude-code-task-costs/method.md), [data](/studies/claude-code-task-costs/results.csv)). The click tasks ran on November 9, 2026 in a library of about 12,000 lines, from the original bug reports, and passed only when the maintainers' own fix tests passed ([method](/studies/codex-vs-claude-code/method.md)). Every Opus 5.5 run passed its check. Bigger repositories and long interactive sessions cost more, because each request re-reads more context. Smaller models cost far less on trivial jobs. On the commit message, Claude Sonnet 5.5 cost $0.034 and Claude Haiku 5.5 $0.0024 per run as Claude Code priced them, and all nine runs made a valid commit; [What common Claude Code tasks cost, measured on each model](/blog/claude-code-task-costs-measured) has the full breakdown. That doesn't carry over to real engineering work: on Terminal-Bench 4.0, Anthropic reported 39.2% for Haiku 5.5 against 70.6% for Sonnet 5.5 ([Anthropic](https://www.anthropic.com/news/claude-haiku-5-5)). Keep bug fixes, refactors and design work on the strongest model you have. At the API-key average of the four click bug fixes, $0.184, Pro's $20 pays for about 110 fixes like them at list prices, Max's $100 about 540 and Ultra's $200 about 1,090. Those are API prices. Anthropic doesn't publish plan allowances in tokens, so outside measurements of what a plan holds, in the next section, are the closest guide. Across enterprise deployments, Anthropic reports Claude Code averaging [about $13 per developer per active day, or $150 to $250 a month](https://code.claude.com/docs/en/costs), at API prices. [Subscription or API: which is cheaper for coding agents?](/blog/subscription-vs-api-pricing) works through the break-even point. ## Are Claude plans subsidized compared with the API? Measured at API list prices, a Claude plan holds many times its price in usage, so for heavy users a plan is by far the cheaper way to run Claude Code. Anthropic doesn't publish plan allowances in dollars, so the figures come from outside tests that watch how far a plan's usage meter moves for a known number of tokens, then price those tokens at API rates: | Who measured, when | Plan | Usage a plan holds, at API prices | | ------------------------------------------------------------------------------------------ | ----------------------------------------------------------- | -------------------------------------- | | [SemiAnalysis](https://semianalysis.com/weekly/episodes/ep-036/), October 7, 2026 | Claude, $200 a month (then Max 20x, now Ultra), on Opus 5.5 | About $12,000 a month | | [Wokey](https://wokey.ai/claude-usage-limits), October 26 to November 2, 2026, 18 accounts | Claude Max 20x (now Ultra), $200 a month | About $1,549 a week, or $6,711 a month | | [Wokey](https://wokey.ai/claude-usage-limits), same week, 79 accounts | Claude Pro, $20 a month | About $155 a week, or $673 a month | The estimates differ because each test assumes its own mix of models and cached tokens: SemiAnalysis tests one token type at a time, and Wokey, which runs a gateway, measures the traffic of real subscription accounts. Both are outside measurements, not figures Anthropic publishes. On the mid-tier models, SemiAnalysis measured [about 5 times the API-equivalent value](https://newsletter.semianalysis.com/p/anthropic-subscriptions-offer-5x) on Claude plans as on ChatGPT plans, comparing Claude Opus 5.5 with GPT-6.1 Sol; Codex pricing explained has OpenAI's side. That value doesn't mean every plan loses money. SemiAnalysis estimates that Anthropic [still books 50% to 60% gross margins on its plans](https://semianalysis.com/weekly/episodes/ep-036/), because API prices carry a high margin and most subscribers use only part of their limit. In its model, a subscriber who uses a whole limit on Opus 5.5 costs Anthropic more than the plan brings in ([SemiAnalysis](https://newsletter.semianalysis.com/p/anthropic-subscriptions-offer-5x)). Anthropic has tightened limits for that kind of use before: it added weekly limits in August 2025 after some subscribers ran Claude Code [around the clock](https://techcrunch.com/2025/07/28/anthropic-unveils-new-rate-limits-to-curb-claude-code-power-users/), and said they would affect fewer than 5% of subscribers. For a developer, the practical point is the break-even: once a month of your Claude Code work at API prices would cost more than a plan, the plan is the cheaper way to pay, and from then on its limits are what you manage. ## How does AutoRouter change what Claude Code costs? AutoRouter, the Mac app for Codex and Claude Code, adds 20+ models to Claude Code and picks the model for each task, so the plan you pay for goes to the work that needs it. In Claude Code, Claude models run on your Claude plan. AutoRouter works out how hard each task is: hard work goes to the most capable model your plans cover, and to credits only when none is strong enough, while everything else goes to the model that costs you least, which is often your plan or a low-cost open model. A model has to clear a quality bar for the task before price counts, and you can see each request's model, what it cost and why. [AutoRouter Free](https://autorouter.com/#plans) costs $0 a month, and routing across your ChatGPT and Claude plans is free. Pro costs $50 a month with $20 of included usage and adds every other model at the provider's price plus 10%; when your Claude plan hits its limit, Pro keeps that work going from the included usage until the plan resets. New accounts get Pro free for their first month, or until $20 of usage is spent. AutoRouter earns the 10% on paid models and nothing on plans. It runs on macOS 14 or later, Claude Code needs Node 22.18 or later, and it is in beta; you choose it with `/model` in Claude Code. More ways to bring the bill down are in [How to cut Claude Code and Codex costs](/blog/cut-claude-code-and-codex-costs). ## Questions ### Is Claude Code free? No. As of November 9, 2026, Claude Code isn't part of Claude's Free plan; it comes with Pro, Max, Ultra, Team and Enterprise, or with a Claude Console account billed per token ([Anthropic](https://claude.com/pricing)). The cheapest way in is Pro at $17 a month billed annually. ### Does Claude Code cost extra on a Claude plan? No. Claude Code is included in every paid plan and shares the plan's usage limits. You pay more only if you turn on [usage credits](https://support.claude.com/en/articles/12429409-manage-usage-credits-for-paid-claude-plans), billed at API rates, or if the `ANTHROPIC_API_KEY` environment variable is set, in which case Claude Code [uses the key instead of your plan](https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan) and bills API rates. ### Can Max or Ultra API credits pay for Claude Code? No. The $100 and $200 of monthly API credits on Max and Ultra are for building on the Claude Platform, and Anthropic says they [don't apply to usage credits in Claude, Claude Code or Claude Cowork](https://support.claude.com/en/articles/12429409-manage-usage-credits-for-paid-claude-plans). ### What happened to Claude Max 20x? Anthropic renamed it Ultra. Existing Max 20x subscribers [moved to Ultra automatically](https://support.claude.com/en/articles/17728952) at the same $200 a month, which now gives 10 times Pro's weekly usage and no five-hour limit for verified accounts. ### Is Claude Code cheaper than Codex? On AutoRouter's November 9, 2026 test of five tasks in the click library, Codex on GPT-6.1 Sol cost $0.085 per task at API list prices, against $0.252 for Claude Code on Opus 5.5 as Claude Code priced it on a Max plan. Claude Code passed all 15 runs and Codex 14. [Codex vs Claude Code](/blog/codex-vs-claude-code) has the full comparison, and Codex pricing explained covers ChatGPT's plans. ## Sources - [Plans and pricing](https://claude.com/pricing), Anthropic, checked November 9, 2026 - [What is the Ultra plan?](https://support.claude.com/en/articles/17728952), Claude Help Center - [What is the Max plan?](https://support.claude.com/en/articles/11049741-what-is-the-max-plan), Claude Help Center - [Use Claude Code with your Pro, Max or Ultra plan](https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan), Claude Help Center - [Manage usage credits for paid Claude plans](https://support.claude.com/en/articles/12429409-manage-usage-credits-for-paid-claude-plans), Claude Help Center - [Pricing](https://platform.claude.com/docs/en/about-claude/pricing), Claude Platform docs, checked November 9, 2026 - [Manage costs effectively](https://code.claude.com/docs/en/costs), Claude Code docs - [Model configuration](https://code.claude.com/docs/en/model-config), Claude Code docs - [How Claude Code uses prompt caching](https://code.claude.com/docs/en/prompt-caching), Claude Code docs - [Introducing Claude Haiku 5.5](https://www.anthropic.com/news/claude-haiku-5-5), Anthropic - [Anthropic Subscriptions Offer 5x+ More Value Than OpenAI](https://newsletter.semianalysis.com/p/anthropic-subscriptions-offer-5x), SemiAnalysis, October 5, 2026 (partly paywalled) - [Ep. 036: $200 Buys $12,000 of Opus Tokens](https://semianalysis.com/weekly/episodes/ep-036/), SemiAnalysis Weekly, October 7, 2026 - [Claude usage limits, measured](https://wokey.ai/claude-usage-limits), Wokey, measured October 26 to November 2, 2026 - [Anthropic unveils new rate limits to curb Claude Code power users](https://techcrunch.com/2025/07/28/anthropic-unveils-new-rate-limits-to-curb-claude-code-power-users/), TechCrunch, July 28, 2025 ## Read next - [How to check your Claude Code usage](https://autorouter.com/blog/check-claude-code-usage.md): Run /usage in Claude Code to see your plan usage bars, a breakdown of what used your plan, and, for API users, a session token and cost estimate. Add a status line to keep plan and context usage on screen. AutoRouter, the Mac app for Codex and Claude Code, also shows the model and cost of every turn. - [What common Claude Code tasks cost, measured on each model](https://autorouter.com/blog/claude-code-task-costs-measured.md): AutoRouter ran five everyday Claude Code tasks three times each on Claude Opus 5.5, Sonnet 5.5 and Haiku 5.5 in October 2026. At Anthropic's list prices, an average run cost $0.110 on Opus 5.5, $0.049 on Sonnet 5.5 and $0.0044 on Haiku 5.5, and all 45 runs passed their checks. Writing the prompt cache was the largest part of every model's cost. - [How to cut Claude Code and Codex costs](https://autorouter.com/blog/cut-claude-code-and-codex-costs.md): Claude Code and Codex costs come down to three things: which model runs each turn, how much context each request carries, and whether you pay through a plan, credits or an API key. Run routine turns on a smaller model, keep sessions short, and spend the plan you already pay for before paying per token. AutoRouter, the Mac app for Codex and Claude Code, makes the model choice on every turn and shows each turn's model, cost and reason. Tags: [Costs](https://autorouter.com/blog/tags/costs.md), [Claude Code](https://autorouter.com/blog/tags/claude-code.md) Drafted with AI from the sources linked above, then checked and edited by the AutoRouter team.