Claude currently costs $0 for Free, $20/month for Pro, $100/month for Max 5x, and $200/month for Max 20x. Pro can also be prepaid annually for $200 per year, which Anthropic displays as roughly $17/month. Team starts at $20 per Standard seat per month on annual billing or $25 monthly, while Enterprise has moved to a $20-per-seat base plus usage billed at API rates.
The important part is no longer just the subscription price. Claude’s paid plans now combine included usage, optional usage credits, Claude Code, Cowork and other products, while some premium models can consume pay-as-you-go credits even on a subscription. That makes the real question: what is included in the plan, what is metered separately, and what kind of user or organization is the plan built for?
AI-XBlog recommendation: Pro is the best-value Claude plan for most individual professionals who use Claude regularly. Max 5x makes sense when Pro limits interrupt paid work; Max 20x is a capacity purchase for very heavy users. Team is for organizations that need central administration and business-data defaults, while Enterprise is now a fundamentally different usage-based buying model rather than simply a more expensive Team tier.
Source check: core plan prices and Claude API refusal/fallback billing were rechecked September 27, 2026; the Claude Agent SDK billing notice was rechecked September 25; broader model/API details and feature packaging were last reviewed September 24. This is a researched living pricing guide based on Anthropic’s pricing page, platform docs, release announcements and Help Center. Claude pricing, included models, usage limits and billing options can change, so verify checkout before purchasing.
September 2026 update: Opus 5.5 changes the cost and capacity picture
Anthropic released Claude Opus 5.5 on September 22, 2026. The direct Claude API price is $4 per million input tokens and $20 per million output tokens, with $0.20/MTok cache reads, $5/MTok 5-minute cache writes, $8/MTok 1-hour cache writes, and 50%-discounted Batch API input/output pricing. Anthropic’s model page lists a 1M-token context window and 128K maximum standard output.
This release does not change the headline Free, Pro, Max, Team, or Enterprise seat prices shown below. It does change the value equation for heavy users: Anthropic says it increased five-hour usage limits on Pro, Max, Team, and seat-based Enterprise plans alongside the launch, and subscription users receive a saveable rate-limit reset. Developers should also note that Opus 5.5 is not always a drop-in replacement for Opus 5 because adaptive thinking is always on and several API/tool-use behaviors changed. For the model-specific migration details, see our Claude Opus 5.5 guide. See Anthropic’s launch announcement and official Opus 5.5 model documentation.
Claude pricing at a glance
| Plan | Current U.S. web price | Best for | Key note |
|---|---|---|---|
| Free | $0 | Occasional Claude use | Limited usage; many core Claude capabilities are available without payment |
| Pro | $20/month or $200/year | Regular individual knowledge and coding work | Annual price equals about $16.67/month; includes Claude Code and broader Claude capabilities |
| Max 5x | $100/month | Frequent power users | 5× Pro session capacity plus higher output limits and priority access |
| Max 20x | $200/month | Very heavy daily users | 20× Pro session capacity; still subject to weekly and other limits |
| Team Standard | $25 monthly or $20/month billed annually | Organizations needing collaboration and administration | Minimum 2 members; more usage per session than Pro |
| Team Premium | $125 monthly or $100/month billed annually | Power users inside Team workspaces | 5× Standard seat usage; Standard and Premium seats can be mixed |
| Enterprise | $20/seat + usage at API rates | Organizations that need enterprise controls and usage-based economics | Annual billing; self-serve or sales-assisted options |
Anthropic says displayed prices are for U.S. customers and exclude applicable taxes. Currency, tax treatment, mobile-app pricing and regional availability can differ.
Pro: $20 monthly or $200 per year
Claude Pro is the simplest paid-plan decision. Anthropic currently charges $20 per month on monthly billing or $200 up front for an annual subscription. The annual plan works out to about $16.67 per month, which Anthropic rounds to $17 in its pricing display.
| Pro billing option | Amount paid | 12-month cost | Savings vs monthly |
|---|---|---|---|
| Monthly | $20/month | $240 | Baseline |
| Annual | $200 upfront | $200 | $40/year |
Pro includes more usage than Free and currently bundles Claude Code, Claude Science, unlimited Projects, Research, access to more Claude models and Claude for Microsoft 365. Anthropic is also rolling the former separate Cowork surface into the unified Claude experience, where Docs, Slides and Design are integrated into the main interface for eligible plans. Anthropic also lists a 200k context window for the individual paid tiers.
Claude Docs, Slides, and the new unified Claude experience
Anthropic made a significant product change on September 16, 2026: Claude chat and Cowork are being merged into a single Claude experience. The rollout starts with Pro and Max over the following weeks, with more plans to follow. In the new experience, you no longer have to decide whether a request belongs in chat or Cowork; Claude routes the task itself.
At the same time, Anthropic launched Claude Docs and Claude Slides in beta on paid plans, while Claude Design now works directly inside normal conversations. Docs can be edited collaboratively and exported to Word, PDF, Markdown, or Google Docs. Slides can be edited and presented inside Claude, then exported to PowerPoint or PDF. Anthropic says Docs, Slides, and Design are on by default for Pro, Max, and Team, while Enterprise owners must enable them; all artifact work counts toward the plan’s normal usage limits.
That makes the paid-plan comparison more important than a simple message-limit chart: Pro and Max are now also access tiers for Anthropic’s expanding knowledge-work surface. Team and Enterprise organizations still retain the existing chat/Cowork setup while the unified experience rolls out more broadly, so availability can differ by workspace and administrator settings.
Official sources: Anthropic: Claude Cowork and chat are now one Claude; Anthropic Help Center: Artifacts, Claude Docs, and Claude Slides; Anthropic Help Center: Claude Docs.
Choose Pro if: Claude is part of your work most days, you want Claude Code or the broader paid feature set, and Free limits are already affecting real tasks. If you know you will use Claude for a full year, the annual plan currently provides a straightforward $40 discount.
Max 5x and Max 20x: buying capacity, not a completely different product
Claude Max comes in two tiers. Max 5x costs $100 per month and Max 20x costs $200 per month on the web. Anthropic describes them primarily as higher-capacity versions of Pro: 5× or 20× more usage per five-hour session, higher output limits, early access to advanced features and priority access during busy periods.
That distinction matters because Max does not mean “unlimited Claude.” Anthropic says Max plans still have a five-hour session reset, weekly limits across models, and potentially other model- or feature-specific caps.
Max pricing over 12 months
| Plan | Monthly price | 12-month equivalent | Difference vs Pro monthly |
|---|---|---|---|
| Pro monthly | $20 | $240 | Baseline |
| Max 5x | $100 | $1,200 | +$960 |
| Max 20x | $200 | $2,400 | +$2,160 |
The jump from Pro to Max is therefore large. Upgrade when usage interruptions have a measurable cost—not just because Max exposes a higher number in the plan selector.
Billing documentation note: Anthropic’s current plan cards, pricing FAQ and August 2026 Max Help Center describe Max 5x and Max 20x as monthly plans. The comparison table on the same pricing page currently lists “Monthly and annual” under the Max billing-cycle row, which conflicts with those statements. Until Anthropic resolves that inconsistency or displays an annual Max checkout price, we treat $100/$200 monthly as the verified public pricing and recommend confirming the checkout screen if you see an annual Max option.
Usage limits are not a fixed message count
Claude pricing is easier to misunderstand if you translate every plan into “messages per day.” Anthropic says usage depends on factors including message length, files, conversation length, model choice, tool usage, effort level and Artifacts. Activity across Claude on web, desktop, mobile and Claude Code draws from the same usage pool on individual paid plans.
The current plan framework is more useful:
- Free: limited everyday capacity.
- Pro: Anthropic says at least 5× more usage per five-hour session than Free.
- Max 5x: 5× Pro session capacity.
- Max 20x: 20× Pro session capacity.
Paid plans can also use optional usage credits to continue beyond included subscription limits at standard API-rate economics. This creates a second cost layer that should be treated separately from the subscription itself.
Claude Code is included—but it shares your plan limits
Claude Code is included with Pro and Max, and Anthropic now includes Claude Code in Team seats as well. This is valuable for developers because one subscription can cover Claude’s chat products and terminal/IDE coding workflows. If you are choosing between coding environments, see our Cursor vs Claude Code in 2026 comparison. If you are evaluating Anthropic’s new coordinator-based cloud workflow, see our Claude Code Projects guide.
But the usage pool is shared for interactive Claude use. Heavy Claude Code sessions can consume the same plan capacity used by Claude conversations. Anthropic also warns that if your environment has an ANTHROPIC_API_KEY configured, Claude Code can authenticate with the API key instead of your subscription, creating separate API charges. For the coding-specific billing breakdown, see our Claude Code Pricing in 2026 guide.
Claude Agent SDK credit: the June change is paused
Do not budget around a separate monthly Claude Agent SDK credit right now. Anthropic’s current Help Center says the previously announced June 15 change was paused. For now, Claude Agent SDK usage, the non-interactive claude -p command and third-party apps using a Claude subscription still draw from the subscription’s normal usage limits, and the previously announced monthly Agent SDK credit is not available.
The older $20 / $100 / $200 monthly-credit table is preserved in Anthropic’s Help Center only as historical reference below the pause notice; it is not the current billing rule. Claude Platform accounts using an API key continue with normal pay-as-you-go API billing. See Anthropic’s current Agent SDK billing notice.
Premium models can add usage-credit costs on top of the subscription
This is one of the most important 2026 pricing details. Anthropic’s current pricing table says Fable models are not included the same way across paid plans:
- Pro: Fable uses pay-as-you-go usage credits.
- Max 5x / 20x: Fable can use up to 50% of weekly included limits before additional credits are needed.
- Team Standard: current Help Center guidance similarly treats Fable as usage-credit based.
- Team Premium: Fable is included as part of the seat’s standard plan usage, with limits.
So a Pro subscription does not guarantee that every premium Claude model is included inside the flat monthly fee. If a specific frontier model is central to your work, check that model’s current plan rules rather than assuming “paid Claude” means unlimited access.
Fable 5.1 follows the same subscription entitlement rules as Fable 5. Its API economics changed materially: Fable 5.1 remains $10/MTok for base input and $50/MTok for output, while cache hits and refreshes are now $0.25/MTok—75% below Fable 5’s $1/MTok cache-read rate. Five-minute cache writes remain $12.50/MTok and one-hour cache writes $20/MTok. Developers can use claude-fable-5-1; Anthropic estimates the lower cache-read price reduces typical workload cost by about 25% and highly agentic workload cost by up to about 45%. See Anthropic’s official API pricing and Fable 5.1 launch details. Anthropic says both models are available on all paid Claude plans, but Pro and Team Standard use pay-as-you-go usage credits from the start. Max and Team Premium can use Fable models inside the normal weekly allowance up to the documented Fable share before extra credits are needed. Importantly, Fable 5.1 was never part of the earlier Fable 5 promotion and does not receive an equivalent one-time promotional credit. Usage-based Enterprise and direct API access are billed at standard API rates.
Team pricing: Standard and Premium seats
Claude Team currently supports organizations of 2 to 150 people and allows Standard and Premium seats in the same workspace.
| Team seat | Monthly billing | Annual billing | Usage relationship |
|---|---|---|---|
| Standard | $25/member/month | $20/member/month | 1.25× Pro session usage |
| Premium | $125/member/month | $100/member/month | 5× Standard seat usage / 6.25× Pro session usage |
Team’s minimum has changed over time. Anthropic’s current Help Center states a minimum of two members. Older articles and older billing documentation that say five seats are no longer a safe basis for a 2026 purchase decision.
Two-person Team cost examples
| Seat mix | Annual-plan monthly equivalent | Annual bill |
|---|---|---|
| 2 Standard | $40/month | $480/year |
| 1 Standard + 1 Premium | $120/month | $1,440/year |
| 2 Premium | $200/month | $2,400/year |
The value of mixed seats is simple: a company can pay for Premium capacity only for the people who actually need it, while keeping the rest of the workspace on Standard seats.
Team is buying governance as well as usage
Team adds centralized billing and administration, SSO, connector controls, organization deployment options, enterprise search and other collaboration features. Anthropic also states that Team content is not used for model training by default.
By contrast, Anthropic’s current individual-plan pricing table lists model training as opt-out for Free, Pro and Max. That does not mean individual content must be used for training; it means the default organizational data-policy posture differs, which can matter when deciding whether a group should use personal subscriptions or a managed Team workspace.
Enterprise pricing changed: $20 per seat plus usage at API rates
Claude Enterprise is no longer only a “contact sales for everything” product. Anthropic’s current pricing page offers a self-serve Enterprise path and lists pricing as $20 per seat plus usage billed at API rates. Enterprise is billed annually.
This means the economic model is different from Team. Team buys included subscription capacity per seat, with optional extra usage credits. Enterprise combines a lower fixed seat charge with usage that scales according to the models and tasks the organization runs.
- SCIM and role-based access
- Audit logs
- Compliance API
- Custom data retention controls
- Network-level access controls and IP allowlisting
- HIPAA-ready offering where applicable
- User and organization spend limits
Sales-assisted Enterprise remains available for organizations that need tailored terms, procurement or a custom commercial arrangement.
Claude subscription vs Claude API: they are not the same bill
Claude subscriptions and Claude API usage are separate products. Pro and Max can include Claude Code when authenticated with your Claude subscription, but direct API use through the Claude Platform is billed by model usage. Enterprise intentionally blends seat and API-rate usage into its pricing model.
API billing caveat for Opus 5.5 and Fable models: a safety-classifier refusal is still a successful HTTP 200 response with stop_reason: "refusal". Anthropic currently bills pre-output refusals in the bio, frontier_llm, and reasoning_extraction categories at the rate of the model that ran the request; other pre-output refusal categories are not billed, although they still count toward rate limits. Mid-stream refusals bill the input plus any output already streamed. If fallback runs, the fallback request is billed separately when applicable, while Anthropic’s fallback credit is designed to avoid paying the prompt-cache setup cost twice. See Anthropic’s refusals and fallback documentation for the current rules.
If you are building automations or agents that call Claude programmatically, budget API tokens and platform features separately from your personal Claude subscription. Our AI Automation in 2026 guide explains why model cost should be treated as its own layer in an automation architecture.
Which Claude plan should you choose?
| If this describes you | Start with | Why |
|---|---|---|
| I use Claude occasionally and do not need Claude Code | Free | Do not pay for capacity you do not use |
| Claude is part of my regular professional or coding workflow | Pro | Best individual value; annual billing reduces 12-month cost |
| Pro limits repeatedly interrupt high-value work | Max 5x | Large capacity increase without paying for the 20x tier |
| I use Claude intensively throughout most workdays and Max 5x is insufficient | Max 20x | Highest individual subscription capacity |
| Two or more people need a centrally managed Claude workspace | Team Standard | Administration, collaboration and organization data defaults |
| A few Team members are much heavier users than everyone else | Mix Premium + Standard | Put 5× seat capacity only where it creates value |
| Usage is large or variable and enterprise controls are required | Enterprise | Seat + API-rate model aligns cost with actual usage |
A practical upgrade test
- Identify the real bottleneck. Are you hitting a session limit, weekly limit, premium-model restriction, output limit or missing organizational control?
- Measure frequency. A limit that appears once a month is different from one that interrupts work every day.
- Separate subscription capacity from usage credits. A higher plan does not automatically make every model or API call free.
- Price the 12-month difference. Pro annual is $200; Max 5x at current monthly pricing is $1,200 over 12 months.
- For teams, decide who actually needs Premium. Mixed seats can be dramatically cheaper than upgrading everyone.
Bottom line
Pro is the rational default paid Claude plan for most individual users. It has the strongest price-to-capability balance, especially at $200/year. Max should be purchased because Pro limits have become a measurable productivity cost, not simply because Max is the highest tier.
For organizations, Team Standard is the clean starting point when central administration and business data handling matter. Use Premium seats selectively for power users. Enterprise becomes compelling when usage is variable enough that seat-plus-API-rate economics and enterprise controls are preferable to fixed included-capacity seats.
If your organization is choosing between the two managed team workspaces, see our ChatGPT Business vs Claude Team comparison. If you are comparing personal ChatGPT and Claude plans more broadly, see our ChatGPT Pricing in 2026 guide. For the broader product decision, see our ChatGPT vs Claude in 2026. For Google’s personal AI plans, see our Gemini Pricing in 2026 guide.
Primary sources
Source check: core plan pricing and refusal/fallback billing rechecked September 27, 2026; broader plan and model packaging remains last fully reviewed September 24–25, 2026. Claude’s plans and model entitlements change frequently; confirm current checkout terms before purchasing.
- Claude official Plans & Pricing
- Claude Help Center: Choose a Claude plan
- Claude Help Center: Max plan
- Claude Help Center: Team plan
- Claude Help Center: Claude Code with Pro or Max
- Claude Help Center: Fable models on subscription plans
- Claude Platform Docs: API pricing and prompt-cache rates
- Anthropic: Claude Fable 5.1 and Mythos 5.1
AI-XBlog Weekly Brief
Keep up with AI that actually works
Join the AI-XBlog Weekly Brief for major AI updates, practical workflows, useful tools, and editor’s picks. No daily noise.
Reader discussion
Join the discussion
Have you tried this tool or workflow? Share your experience, corrections, or questions. Useful reader feedback may help us improve this article.
All comments are reviewed before publication. Your email address will not be published. Promotional links and low-value spam are removed.
