What it is: The complete 2026 pricing breakdown for Claude Code — every plan (Pro, Max, Team Standard, Team Premium, Enterprise), the API/Console pay-as-you-go option, exact rates per million tokens, real-world cost ranges for typical workloads, and how to control spend.
Who it is for: Developers deciding between Claude subscription and API, engineering managers budgeting team rollouts, and anyone trying to predict actual Claude Code costs.
Best if: You want a clear, sourced answer to “what will Claude Code actually cost me?”
Skip if: You haven’t tried Claude Code yet — start with our Claude Code overview and install guide. Subscribe to our free daily newsletter for ongoing pricing changes.
1-on-1 Coaching
Claude AI Crash Course
1-hour private video session with James. Walk through your actual usage and decide which plan fits your work — Pro, Max, Team, or API pay-as-you-go — with concrete numbers from your last week of coding. You leave with a plan choice you can defend to finance.
Group Format
AI Workshops for Engineering Teams
Team-format Claude Code rollouts including plan selection, cost-tracking setup, per-seat allocation, and quota governance. Best for 3+ engineers deciding on Pro vs. Team vs. Enterprise.
What is the bottom line on Claude Code pricing?
Claude Code is included with every paid Claude plan. The simplest mental model: pay $17–$20 per month for Claude Pro and you get Claude Code, Claude Cowork, Skills, MCP, and chat with Sonnet 4.6 and Opus 4.7 bundled together. Heavier users pay $100/month for Max (5× or 20× the usage of Pro). Engineering teams get Claude Team at $20/seat (Standard) or $100/seat (Premium). Enterprise is $20/seat plus metered usage. Or skip subscriptions entirely and pay per-token through the API: $5 / $25 per million tokens for Opus 4.6, less for Sonnet and Haiku.
What are the key takeaways?
- Claude Pro = $17/mo annual or $20/mo monthly. Includes Claude Code for individual developers.
- Claude Max from $100/mo. 5× or 20× more usage than Pro. Right for heavy daily users.
- Claude Team Standard = $20/seat/mo annual, $25 monthly. For teams of 5–150.
- Claude Team Premium = $100/seat/mo annual, $125 monthly. Higher quotas + advanced features.
- Claude Enterprise = $20/seat + metered usage. Custom contracts, SSO, audit logs, HIPAA-ready.
- API (Anthropic Console) = pay-as-you-go. Opus 4.6 at $5/$25 per million tokens.
- Free plan does NOT include Claude Code. You need at least Pro.
How much does Claude Code cost on each plan?
| Plan | Monthly price (annual) | Monthly price (month-to-month) | Claude Code included? | Best for |
|---|---|---|---|---|
| Claude Free | $0 | $0 | No | Trying chat-only Claude |
| Claude Pro | $17/mo | $20/mo | Yes | Individual developers |
| Claude Max (5×) | From $100/mo | From $100/mo | Yes, 5× quota | Daily heavy users |
| Claude Max (20×) | ~$200/mo (verify on pricing page) | ~$200/mo | Yes, 20× quota | Power users running agent teams |
| Claude Team Standard | $20/seat/mo | $25/seat/mo | Yes | Engineering teams 5–150 |
| Claude Team Premium | $100/seat/mo | $125/seat/mo | Yes, higher quota + early access | Larger teams or heavier per-seat usage |
| Claude Enterprise | $20/seat + usage | Custom | Yes | Larger orgs with SSO/audit needs |
| API (Anthropic Console) | Pay-as-you-go | Pay-as-you-go | Yes — metered | Cost-tracking discipline, prod use |
Recommendation for most individual developers: start with Claude Pro at $17/month annual. If you hit limits, upgrade to Max. If you want metered usage instead, switch to Console (API) auth.
What’s included in Claude Pro for Claude Code?
Everything Claude Code needs to function:
- Full Claude Code access across terminal, IDE (VS Code, JetBrains), Slack, web, and CI/CD integrations.
- Access to Sonnet 4.6 and Opus 4.7. Anthropic routes between models automatically based on task complexity.
- Claude Cowork access on the desktop app for non-coding agentic tasks.
- Unlimited projects. No cap on the number of CLAUDE.md projects you can manage.
- Research mode for deeper investigations.
- Microsoft 365 integration.
- Web search built-in.
- Standard rate limits — sufficient for most individual developers shipping a few features per day.
Rate limits on Pro are quota-based, not strict hourly caps. Anthropic publishes guidance like “1,000 messages per week for chat” and “agent runtime sufficient for most daily-driver workloads.” Heavy use can hit the wall by late afternoon on busy days; that’s the signal to upgrade to Max.
When does Claude Max make sense?
Max starts at $100/month and offers two tiers:
- Max 5× — 5 times the usage allowance of Pro. Right for developers who hit Pro limits a few times a week.
- Max 20× — 20 times the usage allowance. Right for daily power users running agent teams, parallel runs, or long agentic sessions.
Beyond the higher quotas, Max plans get priority access during high traffic, early access to new features (Excel/PowerPoint integrations launched in research preview for Max users first), and higher per-task output limits. If you ever wait through “Claude is at capacity” errors on Pro during work hours, Max removes that friction.
Rule of thumb: if you’re shipping more than two non-trivial features per day with Claude Code, Max pays for itself in saved waiting.
What’s the difference between Claude Team Standard and Premium?
Both Team tiers are designed for groups of 5 to 150 people sharing a workspace. The difference is per-seat capacity:
| Feature | Team Standard | Team Premium |
|---|---|---|
| Price (annual) | $20/seat/mo | $100/seat/mo |
| Price (monthly) | $25/seat/mo | $125/seat/mo |
| Claude Code quota per seat | ~Pro-tier | ~Max-tier |
| Workspace shared across team | Yes | Yes |
| SSO | Yes | Yes |
| Admin controls | Yes | Yes + advanced |
| Early access to features | Standard | Yes |
| Higher output limits | No | Yes |
| No training on your content (default) | Yes | Yes |
Standard is the right default. Move to Premium if more than ~30% of your seats are hitting Pro-equivalent quotas or your team needs Max-tier features (parallel agent teams, deeper Excel/PowerPoint integration).
How much does Claude Enterprise cost?
Enterprise pricing is $20/seat plus metered usage. The seat fee covers the workspace, admin tooling, and integrations; the usage cost scales with how much your team actually queries Claude.
Enterprise includes everything from Team plus:
- SCIM-based user provisioning from your identity provider.
- Role-based access controls (RBAC).
- Audit logs for every action.
- Compliance API for SIEM integration.
- Custom data retention policies.
- IP allowlisting.
- HIPAA-ready option (with BAA).
- Claude Security (beta) — AI security tooling for code review and threat modeling.
Pricing is quoted per-customer via the Anthropic sales team. Typical landed cost for a 100-seat enterprise deployment runs significantly less per seat than equivalent Team Premium volume due to negotiated usage rates.
Should you use a Claude subscription or the API for Claude Code?
| Question | Subscription (Pro/Max/Team) | API (Anthropic Console) |
|---|---|---|
| How is it billed? | Flat monthly fee | Per-token, pay-as-you-go |
| Predictable budget? | Yes — you know the bill | No — varies with usage |
| Cost tracking per project? | Limited | Full — every call metered |
| Production / CI/CD usage? | Allowed but not optimal | Right path |
| Best for individual developer? | Yes | Only if you want precise tracking |
| Best for prod workloads? | No | Yes |
| Switch later? | Easy — just /login as API user | Easy — just /login as sub user |
Rule of thumb: subscription for interactive coding, API for automated pipelines. Many developers run both — subscription on their laptop, API for the GitHub Actions workflow that uses Claude Code in CI.
How does Claude API pricing actually work?
The API charges per million tokens of input and output. As of May 2026, published rates:
| Model | Input ($/M tokens) | Output ($/M tokens) | Notes |
|---|---|---|---|
| Claude Opus 4.6 | $5.00 | $25.00 | $10/$37.50 for inputs > 200k tokens |
| Claude Sonnet 4.6 | ~$3.00 | ~$15.00 | Default mid-tier; verify current rate at anthropic.com/pricing |
| Claude Haiku 4.5 | ~$1.00 | ~$5.00 | Fastest, cheapest; verify current rate |
| US-only inference | 1.1× pricing | 1.1× pricing | For data-residency requirements |
For exact, current rates always check anthropic.com/pricing — Anthropic adjusts these periodically and Sonnet/Haiku rates in particular shift more often than Opus.
A token is roughly 0.75 words. So $5 per million input tokens is about $5 per 750,000 words of input — cheap. Output is 5× the price because models produce far fewer output tokens than they consume in input, and output quality is the value users pay for.
What does a typical Claude Code task actually cost on the API?
Real numbers from typical workloads, assuming Opus 4.6 ($5/$25 per million tokens):
- Quick bug fix (read 2-3 files, fix one issue): ~20k input + ~2k output tokens = ~$0.15
- Add a feature (read 5-10 files, write 3 files, test): ~80k input + ~10k output = ~$0.65
- Multi-file refactor (touch 20+ files): ~300k input + ~30k output = ~$2.25
- Codebase audit (read entire mid-size project): ~800k input + ~20k output = ~$4.50 (premium-tier input rate)
- Big agent-team run (parallel agents, hours of work): ~5M input + ~200k output = ~$30–$40
These are rough averages — actual usage depends heavily on your CLAUDE.md content (which loads on every session), how much Claude re-reads files, and how chatty your prompts are. Most individual developers using API auth come in under $30/month even with daily use.
Are there hidden costs in Claude Code?
- Long context premium pricing. Inputs over 200k tokens (using the 1M-context beta) move to $10/$37.50 per million. If you’re feeding whole codebases or large document corpora, expect to pay the premium rate.
- Cloud provider markups. Bedrock, Vertex AI, and Foundry pass through Claude pricing, sometimes with a small platform fee on top. Compare the gross rate against direct Anthropic Console pricing if cost is the deciding factor.
- Output-token surprises. Long-form code generation or large reports can blow up the output token bill. Set
max_tokensin API calls; use/clearin subscription mode to reset stale context. - Re-reading files unnecessarily. If Claude Code re-reads files on every turn, your input bill grows linearly. CLAUDE.md should encode what’s stable so it doesn’t get re-discovered.
- Agent teams in parallel. 5 agents = 5× the token consumption per wall-clock minute. Use agent teams deliberately, not by default.
How do you control Claude Code costs?
- Pick the right plan. Most individuals are over-spending if they’re using API for routine work, or under-served if they’re on Pro for heavy agent work. Match the plan to your actual usage shape.
- Use Sonnet for routine work. Anthropic routes between models, but you can force Sonnet for tasks that don’t need Opus. Sonnet is ~5× cheaper.
- Write a tight CLAUDE.md. Long CLAUDE.md loads on every session. Keep it under ~150 lines; reference linked docs for depth instead.
- Use
/clearbetween unrelated tasks. Don’t drag old context into new conversations. - Use API mode for CI/CD. Production usage should be metered, not subscription-burned.
- Cap
max_tokens. Set a hard ceiling on output tokens for any automated workflow. - Use cost-tracking workspaces. On the Console, first login auto-creates a “Claude Code” workspace specifically for centralized cost tracking. Use it.
Is Claude Code more expensive than Cursor or Copilot?
| Tool | Individual plan price | Model access | Team plan |
|---|---|---|---|
| Claude Code | $17–$20/mo (Pro) | Sonnet 4.6 + Opus 4.7 | $20/seat/mo |
| Cursor | $20/mo (Pro) | Multi-model: Claude, GPT, Gemini | $40/seat/mo (Business) |
| GitHub Copilot | $10/mo (Individual) | GPT + Claude via routing | $19/seat (Business) |
| OpenAI Codex CLI | $20/mo (ChatGPT Plus) | GPT-5.2 | Bundled with Team |
Roughly equivalent on individual pricing. The differentiator is what you get: Claude Code bundles Cowork, Skills, MCP, and the full Claude ecosystem with Claude Code itself. Copilot is cheapest but limited to in-IDE use. Cursor is more expensive per team seat but includes BYOK options for cost optimization.
Frequently asked questions
Is Claude Code free?
No. You need at least Claude Pro ($17/month annual) or an Anthropic Console account with API credits. The free Claude plan does not include Claude Code.
Can I get Claude Code without paying for a Claude subscription?
Yes — sign up for an Anthropic Console account at console.anthropic.com, add credit, and use API-key auth. Pay-as-you-go pricing applies (Opus 4.6 at $5/$25 per million tokens).
Does Claude Pro have unlimited Claude Code?
No — there are quota-based rate limits. Anthropic doesn’t publish exact numbers (they tune them), but most individual developers can ship multiple features per day without hitting limits. Heavy daily use can hit them by late afternoon; that’s the upgrade trigger to Max.
Does the API have a free tier?
The Anthropic Console grants new accounts a small starting credit (~$5) for testing. After that, it’s pay-as-you-go.
Can I deduct Claude Code as a business expense?
If you’re self-employed or own a business, yes — Claude Code is a deductible software/tool expense. Save the invoices from claude.com/billing.
Does Anthropic give educational or non-profit discounts?
Anthropic publishes academic-research credit programs and accepts applications for non-profit access. Contact sales@anthropic.com for details on your specific situation.
What happens if I cancel mid-month?
You keep access through the end of your billing period. After that, Claude Code stops working unless you also have API-key auth configured.
Can I switch from Pro to Max mid-month?
Yes. Anthropic prorates the change. You can also downgrade if you over-bought.
How is Claude Code pricing different from ChatGPT pricing?
ChatGPT Plus ($20/mo) includes Codex CLI for coding. The headline prices are similar; the differences are in model selection (Opus 4.7 vs GPT-5.2), tool ecosystem, and rate limit shapes. See our coding comparison.
Does Anthropic publish historical pricing?
Not formally. Track the Anthropic news page for pricing announcements; Opus 4.6 maintained the $5/$25 pricing of Opus 4.5, but earlier Claude generations had different rates.
Can I share my Claude Pro account with someone else?
Officially no — Pro is a single-user plan. Anthropic detects concurrent unrelated sessions and may suspend accounts that look shared. The right path for multi-person use is Claude Team at $20/seat (still cheaper than buying two Pro plans if both people use it heavily).
What’s the cheapest way to use Claude Code professionally?
Three live options: (1) Claude Pro at $17/month annual covers most individual developers. (2) Pure API mode often comes in under $30/month for daily individual use. (3) For teams, Team Standard at $20/seat with usage tracking is the cheapest path to predictable per-seat cost. Anything less than that and you’re either on the free tier (no Claude Code) or splitting an account (against the terms).
How do I see my current Claude Code usage?
Subscription users: claude.com → Settings → Usage. API/Console users: console.anthropic.com → Usage tab, broken out by workspace. The “Claude Code” workspace that auto-creates on first login is where your Claude Code usage shows up — useful for separating it from other API workloads.
Track every Claude pricing change in one daily email
Anthropic adjusts model pricing, adds tiers, and ships new plan options every few months. The Beginners in AI newsletter ships one issue every day covering what’s new across Claude, ChatGPT, and the rest of the frontier. Free, daily, no fluff.
Sources
- Claude pricing — official subscription tiers
- Anthropic API pricing — per-token rates
- Claude Opus 4.6 pricing details
- Anthropic Console — API access and billing
- Claude models overview — per-model capabilities and rates
- Claude Team plan — details and seat pricing
- Claude Enterprise — SSO, audit logs, compliance
Related news
- Microsoft is canceling Claude Code licenses for thousands of engineers — news, May 14–15 2026.
Want a head start? Book a 2-hour live AI crash course
A private, beginner-friendly session across Claude, ChatGPT, Gemini, Grok, and the wider landscape. Walk away knowing which tools fit your work and how to use them.
Book the 2-hour crash course · $125 →