Claude Code Pricing (2026): Plans, Limits, Ultra

Claude Code doesn't have a price tag of its own. There's no separate "Claude Code plan" to buy. It comes bundled with paid Claude subscriptions, or you can run it on pay-as-you-go API credits from the Claude Console. That makes the real Claude Code cost a little harder to read than a single monthly number.

I use Grok, ChatGPT and Claude every day in my work, so I keep a close eye on how these tools charge for usage. This guide breaks down Anthropic Claude Code pricing for November 2026: every way to pay for it, how the five-hour and weekly limits work, what Ultra adds, what API usage costs, and how it stacks up against Cursor.

All prices were rechecked on November 6, 2026 on Claude's pricing page, Anthropic's API pricing docs, Anthropic's Team plan help article and Cursor's pricing page. They're in US dollars and exclude taxes like GST or VAT. Plans change often, so confirm before you subscribe.

Quick answer: Claude Code is not free. The cheapest paid path is Claude Pro at $20/month, or $17/month billed annually ($200 up front). Pick Max at $100/month when you hit Pro's five-hour or weekly limits most days. Pick Ultra at $200/month for all-day agent work (10x Pro weekly usage, and no five-hour limit for verified accounts, plus $200/month in API credits). For automation, CI or unpredictable spikes, use the API and pay per token. Team Standard seats start at $20/seat/month annually ($25 monthly); Premium is $100/$125. Full Claude chat pricing lives in our Claude pricing guide.

Claude Code pricing at a glance

If you only remember one change from mid-2026 marketing: the old "Max 5x / Max 20x" framing on consumer plans is now Max ($100) and Ultra ($200) on claude.com/pricing. Ultra is the heavy individual tier with the larger weekly pool and the optional removal of the five-hour session cap for verified accounts.

What Claude Code is (and what you actually pay for)

Claude Code is Anthropic's coding agent. It works in your terminal and editor integrations, reads and edits multiple files, runs commands you approve, and handles refactors or bug hunts that would be painful one snippet at a time. On a paid Claude plan it shares the same usage pool as Claude chat on web, desktop and mobile. Your chats and your coding sessions drain one budget.

That shared pool is why Claude Code pricing is really Claude subscription pricing (or API token pricing), not a separate SKU. You're buying:

1. Access — Claude Code is on every paid plan and off Free.
2. Usage headroom — how long you can run before the five-hour or weekly window stops you.
3. Models — Sonnet and Haiku on Free; Opus on Pro and above; Fable with usage credits on Pro and included (up to half your weekly limit) on Max, Ultra and Team Premium.
4. Optional API credits — Max and Ultra include monthly Claude Platform credits; Team orgs can pool up to $500/month.

For how Claude Code feels next to other agents, see Cursor vs Claude Code vs Codex and Claude vs Copilot.

Claude Code pricing plans

Free: no Claude Code

The Free plan covers chat on web, desktop and mobile, web search, Artifacts, memory, connectors and Claude Design / Slides / Docs. Claude Code is not included. You need at least Pro, or an API key from the Console.

Free is still useful for asking coding questions in chat. It is not enough if you want the terminal agent that edits your repo.

Pro: the starting point for Claude Code

Pro costs $20 a month, or $17 a month if you pay $200 for the year up front. Anthropic's pricing page lists everything in Free, plus more usage every five hours and every week, Claude Code, Projects, Research, more models (including Opus), Claude in Chrome and Microsoft 365, hand-off and scheduled tasks, and Claude Dashboards (beta).

Pro is enough for focused coding sessions: fixing bugs, writing small features, explaining unfamiliar code, or a couple of longer agent runs a day. If you live in the agent for several hours every day, you'll likely hit the five-hour or weekly limit and should look at Max.

Annual billing only makes sense if you'll keep the seat for most of a year. Ten months of monthly Pro already cost $200; the annual plan is the same $200 for twelve months.

Max: for daily heavy use

Max is $100 a month, billed monthly only. Everything in Pro, plus:

• 5x Pro usage every five hours and every week
• Fable included for up to half your weekly limit
• $100 a month in API credits on the Claude Platform
• Early access to advanced features and priority access at high traffic times

Max is the plan most full-time Claude Code users land on once Pro's rolling window starts interrupting real work. The included API credits also help if you're prototyping on the Messages API or Managed Agents while you code interactively.

Ultra: for all-day and long-running tasks

Ultra is $200 a month, billed monthly only. Everything in Max, plus:

• No five-hour limit for verified accounts
• 10x Pro usage every week
• $200 a month in API credits (up from $100 on Max)

Ultra is the individual ceiling when Max's weekly pool still isn't enough, or when the five-hour reset keeps stopping long agent sessions. If your account isn't verified, treat the "no five-hour limit" line as conditional and check Anthropic's current footnote on the pricing page before you upgrade for that reason alone.

Team: Standard and Premium seats

Team is for groups of 2 to 150 people. Every seat includes Claude Code, SSO, central billing, admin controls for connectors, project sharing and enterprise search. Anthropic's Team plan help article lists US seat prices as:

• Standard: $25 per seat per month billed monthly, or $20 per seat per month billed annually
• Premium: $125 per seat per month billed monthly, or $100 per seat per month billed annually

You can mix seat types. Premium seats get about 5x the usage of Standard seats (Anthropic's help center frames Standard as about 1.25x Pro per session and Premium as about 6.25x Pro). Team orgs can receive up to $500 per month in pooled API credits on the Claude Platform. Content on Team plans is not used for model training by default.

A five-person team on Standard annual billing is about $100/month before tax, with Claude Code on every seat. That often beats buying five individual Max seats when you also need SSO and shared projects.

Enterprise: seat fee plus API usage

Enterprise is $20 per seat per month billed annually, plus usage billed at API rates. Anthropic's pricing FAQ also notes a larger admin and compliance surface: SCIM, audit logs, custom data retention, role-based access, Compliance API, network controls, IP allowlisting and a HIPAA-ready option. Self-serve and sales-assisted paths both exist; self-serve commonly starts around a 20-seat minimum in Anthropic's published materials. Confirm the current minimum and quote on Anthropic's site or with sales before you budget.

Enterprise is the pick when compliance, invoicing and org-wide spend controls matter more than a flat consumer Max/Ultra fee.

API (Console): pay per token

You can also sign in to Claude Code with an API key from the Claude Console and pay only for the tokens you use. There is no Pro-style five-hour pool. Heavy agentic coding can burn tokens quickly because the agent reads many files and loops on tool results, so set a spending limit early.

How Claude Code usage limits work

This is the part most people misread. According to Anthropic's pricing FAQ on claude.com/pricing:

• Most accounts have a limit that resets on a rolling five-hour session window, and paid plans add weekly limits on top.
• Claude Code, Claude chat on web, desktop and mobile all draw from one shared pool. A long chat session leaves less room for coding, and vice versa.
• There's no fixed message count. Usage depends on conversation length, the model you pick and the features you use.
• Pro gives more usage than Free every five hours and every week (Anthropic describes Pro as at least 5x Free per five-hour session in its FAQ language).
• Max gives 5x Pro every five hours and every week.
• Ultra gives 10x Pro weekly usage, with no five-hour limit for verified accounts.
• On Team plans, Standard seats give more than Pro and Premium seats give about 5x Standard.
• When you hit a limit, you can wait for the reset, upgrade, or on paid plans turn on usage credits to keep working at standard API rates.

You can check where you stand under Settings > Usage.

Practical ways to stretch your limits

• Use Sonnet or Haiku for routine edits and save Opus (and Fable, when included) for hard problems.
• Start fresh sessions for new tasks instead of dragging long context along.
• Point Claude Code at the specific files that matter rather than the whole repository.
• Keep heavy chat work and heavy coding on different days if you're near your weekly cap.
• For CI or batch jobs, move those runs to the API (or Team/Enterprise API credits) so they don't eat your interactive pool.

What's included with Claude Code on paid plans

On Pro and above you get the Claude Code agent plus the rest of the Claude product surface that plan unlocks. Useful pieces for developers:

• Terminal and editor-oriented coding agent with multi-file edits
• Shared Projects and Research on Pro+
• Claude in Chrome and Microsoft 365 on Pro+
• Hand-off and scheduled tasks on Pro+
• Higher-tier models (Opus on Pro+; Fable with credits or included depending on plan)
• Max/Ultra priority access and monthly API credits
• Team/Enterprise admin, SSO and (on Enterprise) compliance tooling

Claude Code is not a separate IDE the way Cursor is. If you want an AI-first editor with inline edits you watch live, compare that workflow in Cursor alternatives and Windsurf vs Cursor. Many people use both: an editor for day-to-day typing, Claude Code for larger agent jobs.

Claude Code with the API: token prices

Instead of (or in addition to) a subscription, you can run Claude Code on Console API credits. Prices below are per million tokens (MTok) from Anthropic's API pricing docs, rechecked November 6, 2026. Model names change often; treat this as the current card, not a permanent rate sheet.

A few API details that change real bills:

• Prompt caching cuts the cost of repeated system prompts and repo context (cache reads are a fraction of base input).
• Batch API is about 50% off input and output for asynchronous work.
• US-only inference (inference_geo: "us") on Claude 4.6+ applies a 1.1x multiplier.
• Managed Agents add $0.08 per session-hour of active runtime on top of tokens.
• Web search on the API is $10 per 1,000 searches, plus token costs.
• Claude 4.7+ and Mythos Preview use a newer tokenizer that can produce about 30% more tokens for the same text than earlier models.

When the API makes sense: light or occasional use, automation in CI pipelines, or teams that want spending tied exactly to usage. When a subscription makes sense: daily interactive coding, where a flat fee is easier to budget and heavy sessions would cost more at token rates. Agentic coding reads a lot of files, so token counts add up faster than most people expect.

If you're new to APIs in general, our guide on what an API is explains how this kind of usage billing works. For no-code automations that call AI models, compare n8n vs Zapier.

Claude Code pricing vs Cursor pricing

Cursor sells plans as one X Subscription that covers Cursor and Grok, with a usage pool that refreshes weekly. We covered it in detail in Cursor Pricing 2026. Here's how the two compare after the November 6, 2026 recheck on cursor.com/pricing:

The big difference: on Cursor, third-party models such as Claude aren't covered by the monthly fee. If you mainly want Claude models, a Claude subscription includes them, while Cursor would charge for them by usage on top of Premium and above.

The other difference is workflow. Claude Code is an agent that works in your terminal and IDE integrations and handles multi-file tasks that you review afterward. Cursor is a full AI code editor where you watch and approve changes as they happen. For a deeper look at which fits which kind of work, read Cursor vs Claude Code vs Codex. If you're shopping editors, Cursor alternatives lists free and paid options.

Which Claude Code plan should you choose?

Your Claude Code cost per month comes down to how often you code with the agent and whether you also want Claude for writing and research in the same seat.

• Trying it out or coding a few times a week: Pro at $20/month, or $17/month annually if you're sure you'll stick with it.
• Coding with an agent most of the day: Max at $100/month.
• Running long agent sessions back to back, or constantly hitting Max's weekly pool: Ultra at $200/month.
• Automation, CI, or unpredictable usage: the API with a spending limit.
• A company team that needs SSO and admin controls: Team Standard for most seats, Premium for power users, or Enterprise when compliance and custom retention matter.

Start with Pro on a real project for a couple of weeks and watch Settings > Usage. If you keep hitting the five-hour limit, that's your signal to move to Max. If Max's weekly cap still interrupts shipping weeks, look at Ultra or Team Premium before you assume the API will be cheaper.

Not sure whether you need Claude for coding at all? Our ChatGPT vs Claude comparison covers everyday writing, research and business use. For Claude vs Google's assistant on writing and coding, see Claude vs Gemini. For every Claude plan beyond coding, including Ultra, Team and per-token API rates, see the full Claude pricing guide.

How to migrate or switch without wasting a month

• Pro → Max/Ultra: upgrades take effect right away; unused time on the old plan is credited toward the new one per Anthropic's billing FAQ.
• Monthly → annual (Pro or Team): remaining monthly time is credited to the first annual invoice.
• Downgrades: take effect at the end of the current billing period; chats, projects and files stay with the account.
• Leaving Cursor for Claude Code (or running both): export or note your Cursor rules/skills you care about, then recreate the important ones as Claude project instructions or repo docs. Keep Cursor on Premium only if you still want the editor UX; don't pay for Claude models twice without checking usage.
• Moving a team: start with Team Standard seats for everyone, upgrade a few developers to Premium after you see Settings > Usage for a week, instead of buying Premium for the whole org on day one.
• Cancel: cancel at least 24 hours before renewal. Web/desktop cancels in Settings > Billing; App Store / Google Play subscriptions cancel in those stores.

Privacy note

Anthropic's consumer plans and Team/Enterprise plans are not the same on training and admin controls. Team and Enterprise content is not used for model training by default, and Enterprise adds compliance-oriented surfaces (audit logs, Compliance API, custom retention, HIPAA-ready option). On consumer plans, check the current opt-out and data-use terms for your exact plan before you connect private repos or paste client code. Cursor's privacy mode (when enabled) is a separate promise on the Cursor side — don't assume one product's setting covers the other if you use both.

If you handle regulated data, read Anthropic's current terms and security pages for your plan rather than relying on a blog summary.

If you call Claude from your own code instead of through a plan, token rates are in Claude API pricing, and which model suits which job is in Claude models explained.

FAQ