Claude Pricing Explained: Free, Pro, Max and API Costs in 2026
By
Alexander Gran
Founder & Editor
This breakdown is based on official documentation from anthropic.com and support.claude.com, as of August 2026. Anthropic changes pricing and limits without much notice — verify current numbers on claude.ai/upgrade before subscribing.
Claude Pro at $20/month is the best choice for most individual users. Max is worth it only for people who run Claude — especially Claude Code — heavily every single day and regularly hit Pro’s limits. The API is billed completely separately, by the token, and can quietly become the most expensive way to use Claude if you’re not watching usage.
Claude Plans at a Glance
| Plan | Price | Best for |
|---|---|---|
| Free | $0 | Occasional chat, testing Claude before paying |
| Pro | $20/mo ($17/mo billed annually) | Most individual daily users |
| Max 5x | $100/mo | Heavy daily users who outgrow Pro |
| Max 20x | $200/mo | Power users running Claude Code most of the day |
| Team | $20–25/seat/mo (Standard), $100–125/seat/mo (Premium) | Small teams, 2–150 seats |
| Enterprise | $20/seat/mo + API-rate usage | Large organizations needing SSO, SCIM, audit logs |
| API | Pay per token — $1–$5 per million input tokens, $5–$25 per million output | Developers building apps, automated pipelines |
Free
Claude’s free tier is more capable than most competitors’ free tiers: according to Anthropic’s pricing page, it includes chat on web, iOS, Android, and desktop; code generation and data visualization; web search; memory across conversations; file creation and code execution; and even extended thinking for complex problems. What Free does not include is Claude Code, Claude Cowork, unlimited projects, or access to Anthropic’s full model lineup — those are Pro-and-up features.
Usage resets on a rolling basis, and Anthropic doesn’t publish a fixed message count — actual limits depend on message length, conversation length, and which model you’re using. Free is genuinely useful for occasional questions, but anyone using Claude daily will hit its ceiling quickly.
Pro
Pro is $20/month billed monthly, or $17/month ($200 billed upfront) on an annual subscription, per Anthropic’s official pricing page. It includes everything in Free, plus Claude Code, Claude Cowork, Claude Design, Claude Science, unlimited projects, access to Research, and access to more Claude models. Anthropic states Pro provides “at least five times the usage per session compared to our free service,” with usage limits that reset every five hours plus a separate weekly cap.
The most important thing to know about Pro: Claude Code is included in the same subscription, not billed separately. Per Anthropic’s support documentation, “Claude Code and Claude share the same usage limits” — everything you do in the web app, mobile app, desktop app, and terminal draws from one shared pool, not separate quotas per surface.
Max 5x
Max 5x costs $100/month and gives five times the usage per session that Pro gives, according to Anthropic’s Help Center. Everything else — model access, features, context window — matches Pro; the difference is purely the size of the usage pool, plus higher output limits, early access to new features, and priority access during high-traffic periods.
Max 5x is the right upgrade if you’re on Pro and consistently hitting your five-hour session limit or weekly cap — not if you’re just curious what more expensive gets you.
Max 20x
Max 20x costs $200/month and gives twenty times Pro’s per-session usage. This tier exists for people running Claude Code most of the working day, or juggling long agentic sessions and large codebases where Max 5x’s larger-but-still-finite pool still runs out. It’s a 2x price jump over Max 5x for 4x the usage headroom on paper, so it only pencils out if you’re actually usage-constrained on Max 5x — not as a default “get the best plan” purchase.
Team and Enterprise, Briefly
Team is built for 2–150 seats: a Standard seat is $20/seat/month billed annually ($25 monthly) with more usage than Pro, and a Premium seat is $100/seat/month billed annually ($125 monthly) with five times a Standard seat’s usage — plus Claude Code, Claude Cowork, Claude Design, and Claude Science included, and connectors to Microsoft 365 and enterprise search.
Enterprise is built for large organizations: $20 per seat per month billed annually, plus usage billed at API rates on top — so total cost scales with how much your team actually runs, not a flat per-seat number. It adds SSO, SCIM, audit logs, custom data retention, and role-based access on top of everything in Team.
API Pricing, Separately
The API is billed per token and has nothing to do with your claude.ai subscription — a Pro or Max plan does not include any API credit. Current per-million-token rates, per Anthropic’s published pricing:
| Model | Input | Output |
|---|---|---|
| Claude Opus 5 | $5.00 | $25.00 |
| Claude Sonnet 5 | $3.00 ($2.00 intro price through 2026-08-31) | $15.00 ($10.00 intro) |
| Claude Haiku 4.5 | $1.00 | $5.00 |
Two levers cut real API costs: prompt caching (cached reads run at roughly 0.1x the base input price once a prompt prefix is reused) and the Batch API (50% off standard token prices for non-latency-sensitive jobs). Neither is automatic — you have to structure requests to use them. Without either, a chat-heavy API integration on Sonnet can rack up cost fast, since every message resends the full conversation history at full price.
Claude Code: What’s Included vs. What Costs Extra
This is where the most expensive surprise hides. Claude Code works two different ways depending on how you authenticate:
- Logged into a Pro or Max subscription: Claude Code usage draws from the same shared pool as your chat usage — no separate charge, no separate limit. This covers the terminal CLI and IDE integrations for VS Code, Cursor, other VS Code forks, and JetBrains IDEs.
- With an
ANTHROPIC_API_KEYenvironment variable set: Claude Code switches to pay-per-token API billing instead of your subscription, per Anthropic’s support documentation. This happens automatically if the key is present — it’s not a prompt you’ll necessarily notice.
If you want Claude Code to stay inside your Pro or Max subscription, Anthropic’s own guidance is to decline any API credit offer it presents and let your usage period reset naturally rather than continuing on pay-per-token billing. If you’re not sure whether Claude Code is drawing from your subscription or your API key, check whether ANTHROPIC_API_KEY is set in your shell environment — that’s the switch.
If you’re deciding whether Claude Code fits your workflow at all, see Cursor vs. Claude Code for how it compares to an IDE-based alternative, or how to choose an AI coding agent for a broader framework before committing to any tool.
Real Spending Scenarios
A solo developer using Claude Code a few hours a day for a side project, plus occasional chat for research, comfortably fits inside Pro’s $20/month — the shared usage pool covers both without hitting the five-hour session cap regularly.
A full-time developer running Claude Code most of the workday, with long agentic sessions across a large codebase, is the profile Max 5x or Max 20x is built for — Pro’s session limits become a real interruption at that usage level, and $100–200/month is still cheaper than the same volume of usage billed at raw API rates.
A team building a customer-facing AI feature on top of Claude — a chatbot, a document processor, an automated pipeline — needs the API, not a claude.ai plan. At Sonnet 5’s $3/$15 per-million rate, a feature processing even a modest volume of requests with full conversation history resent each turn can outpace a $20 Pro subscription within days, which is why API costs need active monitoring (token counting, caching, and the Batch API) rather than being treated like a flat subscription.
Who Should Choose Which Plan
- Free — occasional users who chat with Claude a few times a week and don’t need Claude Code
- Pro — most individual users, including solo developers using Claude Code part of the day
- Max 5x — daily heavy users who consistently hit Pro’s session or weekly limits
- Max 20x — people running Claude Code as their primary daily tool for most working hours
- Team/Enterprise — organizations needing centralized billing, admin controls, or compliance features
- API — anyone building a product or automated workflow on top of Claude, not chatting with it directly
When Max Is Not Worth It
Max is a bad purchase in three common situations: if you’re buying it speculatively (“just in case I need more”) rather than because you’ve actually hit Pro’s limits; if your usage is bursty rather than sustained — a few heavy days a month don’t justify paying for 5x or 20x capacity every month; or if what you actually need is API access for a specific integration, in which case Max’s subscription usage pool doesn’t apply at all and you’re paying for the wrong thing. In all three cases, staying on Pro and upgrading only after you’ve hit a real, repeated limit is the cheaper path.
FAQ
Does the Pro or Max plan include API access?
No. Claude.ai subscriptions (Free, Pro, Max, Team) are separate from the API. Using the API — for your own app, script, or integration — requires a separate Anthropic Console account and is billed per token regardless of any claude.ai subscription you hold.
Is Claude Code free with a Pro subscription?
Yes, as long as you’re authenticated via your Pro or Max login and don’t have an ANTHROPIC_API_KEY set. Claude Code usage then draws from the same shared session and weekly limits as your regular Claude chat — there’s no extra charge.
What’s the actual difference between Max 5x and Max 20x?
Usage pool size only — 5x vs. 20x Pro’s per-session usage. Features, model access, and context window are otherwise the same across Pro and both Max tiers, per Anthropic’s plan comparison.
How much cheaper is annual billing?
On Pro, annual billing brings the effective price to $17/month ($200 billed upfront) versus $20/month billed monthly — about a 15% discount. Team’s Standard seat shows a similar gap ($20/mo annual vs. $25/mo monthly).
Can I switch from Pro to Max mid-cycle?
Yes. Anthropic prorates the switch and credits any remaining value from your old plan toward the new one; if that credit exceeds one billing period of the new plan, the excess is stored and applied to future charges.
Final Verdict
For most people, Claude Pro at $20/month is the right starting point — it already includes Claude Code, and its usage pool covers typical daily use without friction. Upgrade to Max only after you’ve actually hit Pro’s limits repeatedly, not preemptively, and pick 5x or 20x based on how close to constant your Claude Code usage really is. If what you’re building is a product or integration rather than a personal workflow, skip the subscriptions entirely and go straight to the API — but budget for it actively, since per-token billing has no fixed monthly ceiling the way a subscription does.