Claude Code Pricing: A Full Comparison of Free, Pro, Max, and Team Plans

A deep comparison of Claude Code's pricing tiers — what the free tier actually covers, how Pro, Max, and Team differ, the right plan from individuals to teams, and a cost-benefit analysis from the perspective of an IT company founder.
代表 / エンジニア
"I want to try Claude Code, but the pricing is confusing — which plan should I pick?" — this is the most common question I get from engineers and decision-makers looking at adopting it.
Claude Code has several plans (Pro, Max, Team, Enterprise), plus an option for metered API usage. This article lays out each plan's pricing, features, and usage limits, and walks through picking the right plan by team size — from solo developers to companies.
I run an IT company and use Claude Code heavily in my own day-to-day development. Drawing on that experience, I'll put concrete numbers behind "which plan is actually the best deal."
Claude Code Pricing Plans (April 2026)
Here's a snapshot of Claude Code pricing as of April 2026:
| Plan | Monthly | Claude Code | Usage guideline (5-hour window) | Target |
|---|---|---|---|---|
| Free | Free | Not available | — | Chat only |
| Pro | $20/month ($17/month annual) | Available | ~45 messages | Individual developers |
| Max 5x | $100/month | Available | ~225 messages | Daily users |
| Max 20x | $200/month | Available | ~900 messages | All-day power users |
| Team Standard | $25/seat/month ($20 annual) | Not available | — | Chat-only team usage |
| Team Premium | $125/seat/month ($100 annual) | Available | Equivalent to Max 5x | Team development |
| Enterprise | Contact sales | Available | Custom | Large organizations |
| API metered | Pay as you go | Available | Unlimited | Automation, CI/CD |
First thing to internalize: the free plan does not include Claude Code. You need at least Pro ($20/month) to use Claude Code.
Usage limits are managed on a "rolling 5-hour window" — not per-day limits, but a resetting window of the last 5 hours of usage. Use it all up in the morning, and you'll be good again by early afternoon.
Trying Claude Code for Free
Claude Code has no free tier, but there are ways to effectively try it for free.
API Free Credits
New Anthropic API signups get $5 of free credits. You can use these to try Claude Code via the API.
How far does $5 go? For Sonnet 4.6 ($3 input / $15 output per million tokens), assuming a 1:1 input-to-output ratio, it's roughly 550K tokens — enough to edit a medium-sized file 10-15 times. Good for a "see what Claude Code feels like" tour; nowhere near enough for serious development, which will burn through it in hours.
Pro Plan Trial for One Month
The approach I most recommend is signing up for Pro ($20/month) for just one month. No commitment beyond that — cancel anytime, and skipping the annual plan means no early-termination penalty.
In my case, one week was enough to decide whether Claude Code fit my development style. If it doesn't, you cancel at the end of the month and you're out $20.
Pro Plan ($20/month) in Detail
Pro is the cheapest plan that includes Claude Code.
Pricing
- Monthly: $20/month
- Annual: $17/month ($200 per year). Saves $36/year versus monthly billing
Usage Limits
About 45 messages per 5-hour rolling window. Both Sonnet 4.6 and Opus 4.6 are available, and token consumption varies by model.
What does 45 messages actually get you? Some concrete scenarios:
- Bug fixing: Paste an error log, ask "fix this bug" → get a proposed fix → review → apply. 3-5 messages per issue. At 45 messages: 9-15 bug fixes
- Code review: Feed it a PR diff, ask for review → discuss findings. 2-3 messages per PR. At 45 messages: 15-20 PR reviews
- New feature implementation: Describe requirements → design proposal → implementation → write tests → iteration. Typically 10-15 messages for a medium-sized feature. At 45 messages: 3-4 features
Who Pro Is Right For
- Individual developers trying Claude Code for the first time
- People who use AI assistants a few times a week for specific tasks
- People focused on personal projects or OSS contributions
- People who want to start small and decide later
Caveats with Pro
When you hit the 5-hour window limit, you get temporarily rate-limited. Large-scale refactors (changes spanning dozens of files) or long pair-programming-style sessions will run into this.
New since March 2026: "Extra Usage" was introduced. When you exceed the cap, you can auto-switch to metered billing at standard API rates. You can set a monthly cap to prevent surprises.
Max 5x Plan ($100/month) in Detail
5x the usage of Pro. Aimed at developers who use Claude Code daily but not all day long.
Usage
About 225 messages per 5-hour window — 5x of Pro.
What 225 messages actually looks like:
- A typical day: 5 bug fixes in the morning (25 messages) → 10 PR reviews over lunch (30 messages) → 2 feature implementations in the afternoon (30 messages). Total: 85 messages, 38% of the 5-hour window. Comfortable headroom
- A focused coding day: Even if a single large refactor eats up 100 messages, you still have 125 left
Who Max 5x Is Right For
- People who develop alongside Claude Code for 3-4 hours a day
- Pro users who hit rate limits 2-3 times a week
- Freelancers juggling multiple projects in parallel
- People ready to promote Claude Code from "a handy tool" to "a primary development partner"
Breakeven vs. Pro
"Pro isn't quite enough — should I move up to Max 5x?" The criterion is clear.
If you're hitting Pro's rate limits twice a week or more, move to Max 5x. Each limit hit costs you 30 minutes to 2 hours of waiting, times 8+ times per month = 4-16 hours of lost time monthly. At $30/hour for engineering time, that's $120-$480 of monthly opportunity cost. The $80 delta (Max $100 vs Pro $20) pays for itself immediately.
Max 20x Plan ($200/month) in Detail
20x the usage of Pro. For power users who center their development around Claude Code and use it all day.
Usage
About 900 messages per 5-hour window. Effectively "unlimited" for all practical purposes. Independent tests rarely report Max 20x users actually hitting the cap during normal daily use.
Max 20x-Exclusive Perks
From March 2026, Max 20x users get a 1M-token context window when using Opus 4.6 — automatically, at no extra cost, with no extra setup.
This unlocks large refactors that ingest an entire giant monorepo, or architectural changes that require a bird's-eye view of the whole project.
Real Cost-Benefit
One user reported burning through about 10 billion tokens via Claude Code over eight months. On metered API (Sonnet 4.6, $3 in / $15 out per million tokens), that's over $15,000. Eight months of Max 20x is $1,600 — about 90% savings.
At aduce we've standardized on Max 20x as our development environment. $200/month (around ¥30,000) looks steep at first glance, but it's on par with a single engineer's daily cost. Given the productivity gains Claude Code brings, it's easily justified.
Who Max 20x Is Right For
- People who use Claude Code as their primary development tool
- People who've adopted agentic coding as standard
- People doing frequent large-scale codebase operations (migrations, refactors)
- People who want to take advantage of the 1M-token context window
Team Plan in Detail
The plan for companies and teams adopting Claude Code. Minimum 5 seats.
Seat Types and Pricing
| Seat type | Annual | Monthly | Claude Code | Context |
|---|---|---|---|---|
| Standard | $20/seat/month | $25/seat/month | Not available | 200K |
| Premium | $100/seat/month | $125/seat/month | Available | 200K |
Mixing Standard and Premium in the same team is allowed — this is the biggest cost optimization lever.
Cost Examples by Team Composition
Startup (5 people: 3 engineers + 2 non-engineers)
| Seat type | Count | Monthly | Subtotal |
|---|---|---|---|
| Premium | 3 | $125 | $375 |
| Standard | 2 | $25 | $50 |
| Total | 5 | $425/month |
Compared to everyone individually on Max 20x ($200 × 5 = $1,000/month): 57% savings. Non-engineers don't need Claude Code, so Standard is enough for them.
Mid-sized team (15 people: 8 engineers + 2 PMs + 5 others)
| Seat type | Count | Monthly | Subtotal |
|---|---|---|---|
| Premium | 8 | $100 (annual) | $800 |
| Standard | 7 | $20 (annual) | $140 |
| Total | 15 | $940/month |
$11,280 per year. Compared to a single engineer's annual salary ($60,000-$100,000), AI-assisted development for the entire team costs $11,280/year.
Additional Team Plan Features
- Team management: Add/remove members and change seat types from an admin console
- Consolidated billing: Unified corporate billing instead of personal cards
- Security: Organization-level access control
Enterprise Plan in Detail
A custom plan for large organizations. Pricing is negotiated directly with Anthropic's sales team.
Enterprise-Only Features
- 500K-token context window (Team/Max Pro are 200K)
- SSO: Okta, Azure AD, SAML 2.0 supported
- SCIM: Automatic user provisioning and de-provisioning
- Audit logs: Record of who generated what and when
- Custom data retention: Customize data retention periods
- HIPAA compliance: For projects handling healthcare data
- IP address restrictions: Only allow access from specific IPs
- Role-based access control: Separate admin, member, and viewer permissions
When You Need Enterprise
- Engineering organizations with 100+ employees
- SOC 2, HIPAA, ISMAP, or similar compliance requirements
- Need to integrate with an internal identity system (Okta, etc.)
- Regulated industries that require audit logs (finance, healthcare, public sector)
API Metered Billing in Detail
Separate from subscription plans, you can also use Claude Code via the API. Appropriate for automation and CI/CD integration.
Pricing for Major Models
| Model | Input | Output | Notes |
|---|---|---|---|
| Opus 4.6 | $5/1M tokens | $25/1M tokens | Highest performance. For complex reasoning |
| Sonnet 4.6 | $3/1M tokens | $15/1M tokens | Balanced. Best for day-to-day development |
| Haiku 4.5 | $1/1M tokens | $5/1M tokens | Fast and cheap. For simple tasks |
Cost-Reduction Techniques
API billing costs can be slashed with the right techniques.
Prompt caching: When you reuse the same system prompt repeatedly, enabling caching drops reads to 10% of standard cost. A 5-minute cache write costs 1.25x; a 1-hour cache, 2x. Read it twice or more and you're ahead.
Batch API: For workloads where instant response isn't required (bulk code reviews, test case generation), the Batch API gives you a 50% discount on both input and output.
Combining caching and batching: Together, up to 95% cost reduction is possible.
A Concrete Cost Calculation
Scenario: Auto-review 50 PRs per day in CI/CD
- Per PR input: ~5,000 tokens (diff + system prompt)
- Per PR output: ~2,000 tokens (review comments)
- Model: Sonnet 4.6
- Monthly: 50 PRs × 20 business days = 1,000 PRs
| Item | Calculation | Cost |
|---|---|---|
| Input | 5,000 × 1,000 × $3/1M | $15.00 |
| Output | 2,000 × 1,000 × $15/1M | $30.00 |
| Monthly total | $45.00 |
With prompt caching, the system prompt portion (~3,000 tokens) drops to $0.90/1M for cache reads, cutting the monthly cost down to about $25.
Subscription vs. API: Which Is Better?
| Monthly usage (approx. tokens) | Recommendation | Reason |
|---|---|---|
| Under 1M tokens | API | Less than $20 |
| 1M-5M tokens | Pro ($20) | API would be $20-$100 |
| 5M-20M tokens | Max 5x ($100) | API would be $100-$400 |
| 20M+ tokens | Max 20x ($200) | API would be $400+. Max is overwhelmingly better |
Cost-Benefit by Plan
A holistic comparison of annual cost and expected productivity gains.
| Plan | Annual cost | Per day | Expected usage/day | Monthly message estimate |
|---|---|---|---|---|
| Pro | $240 | ~$0.8 | 1-2 hours | ~1,350 |
| Max 5x | $1,200 | ~$3.3 | 3-4 hours | ~6,750 |
| Max 20x | $2,400 | ~$6.6 | 6+ hours | ~27,000 |
| Team Premium (1 seat) | $1,200-$1,500 (annual/monthly) | — | Team use | Equivalent to Max 5x |
Compared to a Full-Time Engineer's Salary
Annual cost of hiring a full-time engineer:
- Salary: ~$40,000-$70,000
- Benefits and overhead: ~15% of salary
- Total: ~$46,000-$80,000
Annual cost of Claude Code Max 20x: $2,400. About 5-8% of a full-time hire.
Of course, Claude Code doesn't replace a full-time engineer. But viewed as a tool that doubles or triples existing engineers' productivity, a $2,400 annual investment is extremely reasonable.
Recommended Plans by Size
Individuals and Freelancers
Recommendation: Pro → Max 5x as needed
Start with Pro for one month. If you hit rate limits twice a week or more, move to Max 5x. $20-$100/month for a meaningful productivity boost as a freelancer.
Example: A freelancer at $50/hour who saves an hour of work per day with Claude Code captures $1,000 of monthly time value (20 business days). ROI on Max 5x ($100/month) is 10x.
Startups and Small Teams (5-10 people)
Recommendation: Team (mix of Standard and Premium)
Give engineers Premium seats and non-engineers Standard seats. A 5-person team (3 engineers + 2 non-engineers) gets everyone using Claude (either code or chat) for $425/month.
Mid-sized Companies (10-50 people)
Recommendation: Team Premium → evaluate Enterprise
Team Premium is the baseline for engineering teams of 10+. Consider moving to Enterprise when SSO, SCIM, and audit logs become requirements.
Trigger for considering Enterprise: When IT or leadership asks "we need visibility into who's generating what with AI."
CI/CD and Automation
Recommendation: API metered billing
For GitHub Actions and CI/CD pipeline integration, API is the only sensible choice. Combine prompt caching and the Batch API to optimize costs. The PR auto-review example above runs for $25-$45/month.
"When in Doubt" Flowchart
Q1: New to Claude Code?
├── Yes → Try Pro ($20/month) for one month
└── No
└── Q2: Hitting rate limits 2+ times per week?
├── No → Stay on Pro
└── Yes
└── Q3: Using it all day (6+ hours)?
├── No → Max 5x ($100/month)
└── Yes → Max 20x ($200/month)
For teams:
Q4: Using it as a team?
├── Yes → Team (mix of Standard + Premium)
│ └── Q5: Need SSO or audit logs?
│ ├── No → Stay on Team
│ └── Yes → Enterprise
└── No → Individual plans (see flow above)Frequently Asked Questions
Can I change plans at any time?
Yes. Pro → Max, Max 5x → Max 20x — all changes are instant, and prorated for the difference. Downgrades work the same way.
How does Max 5x differ from Team Premium?
Usage is roughly equivalent (5x). The difference is management: Team Premium adds consolidated billing, member management, and organization-level security settings. For individuals, Max is appropriate; for multiple people, Team Premium.
Can I combine metered API with a subscription?
Yes. For example: Max 20x for daily development work, metered API for CI/CD automation. API is contracted separately through Anthropic Console.
Should I pay annually?
Pro's annual saves $3/month ($36/year). Max and Team Premium both offer annual discounts — Team Premium in particular ($125 monthly → $100 annual, $300/seat/year difference) scales with headcount.
Summary
For an overview of Claude Code and how to install it, see What Is Claude Code? The Complete Guide.
Picking a Claude Code plan comes down to usage frequency and purpose:
| Criterion | Recommended plan | Monthly |
|---|---|---|
| Just want to try it | Pro | $20 |
| Using it daily | Max 5x | $100 |
| Using it all day / primary tool | Max 20x | $200 |
| Team adoption | Team Premium | $100-$125/seat |
| CI/CD and automation | API metered | Pay as you go |
| Governance requirements | Enterprise | Contact sales |
The most important point is to start small and scale up based on usage. Try Pro for one month to confirm Claude Code fits your development style — that's the most rational way to begin.
At aduce Inc., we provide comprehensive IT advisory support — from adopting AI tools including Claude Code to lifting development team productivity. If you're looking to adopt Claude Code but are unsure about plan selection or operational design, please feel free to reach out via Contact.
References
- Plans & Pricing | Claude by Anthropic
- Pricing - Claude API Docs
- Claude Max Plan Explained | IntuitionLabs
- Claude Code Pricing in 2026: Every Plan Explained - SSD Nodes
※ Pricing is as of April 2026. For the latest pricing, please check the Anthropic official site.