Yes, the $20 Pro plan includes Claude Code. Anthropic's pricing page lists "Includes Claude Code" on Pro, which costs $17/month billed annually or $20 month-to-month. The free plan does not include Claude Code at all.
Below: every plan price, how the 5-hour session and weekly caps actually work, exact per-token API rates with prompt caching math, the breakeven where a Max subscription beats pay-as-you-go, and the June 15, 2026 Agent SDK credit change. All numbers verified against claude.com/pricing and platform.claude.com docs on June 9, 2026.
Every Claude Plan in One Table
| Plan | Price | Claude Code | Usage |
|---|---|---|---|
| Free | $0 | Not included | Chat only |
| Pro | $17/mo annual, $20 monthly | Included | 5-hour session limit + weekly caps, shared with chat |
| Max 5x | $100/mo | Included | 5x more usage per session than Pro |
| Max 20x | $200/mo | Included | 20x more usage per session than Pro |
| Team standard | $20/seat annual, $25 monthly | Included | Per-seat limits |
| Team premium | $100/seat annual, $125 monthly | Included, plus Claude Cowork | Higher per-seat limits |
| Enterprise | Negotiated seat price | Included | Seat price + usage billed at API rates |
| API key (Console) | Pay per token | Works with any Console account | Sonnet $3/$15, Opus $5/$25 per MTok |
Shared usage pool
On Pro and Max, Claude Code and Claude.ai chat draw from the same quota: one five-hour session limit plus weekly limits. Heavy chat use during the day leaves fewer Claude Code turns in the evening. Run /usage or /status inside Claude Code to see where you stand. Full mechanics in our Claude Code usage limits guide.
Does the $20 Pro Plan Include Claude Code?
Yes. The $20 Pro plan includes Claude Code. The line item on Anthropic's pricing page reads "Includes Claude Code", and the setup docs list Pro as the minimum subscription tier: "The free Claude.ai plan does not include Claude Code access." Annual billing drops Pro to $17/month.
What Pro buys you in practice: Claude Code in the terminal, VS Code, JetBrains, and the desktop app, with usage metered against the plan's five-hour session limit and weekly caps, shared with Claude.ai chat. Anthropic stopped publishing fixed messages-per-window or hours-per-week figures, so any page quoting "45 messages per 5 hours" is repeating stale 2025 numbers. The honest answer is: Pro covers light daily coding; if you hit the session cap most days, that is the signal to move to Max 5x.
Claude Code also runs on Amazon Bedrock, Google Vertex AI, and Microsoft Foundry with no Anthropic subscription, billed through your cloud provider. Setup details in our Claude Code subscription guide.
How Limits Work: 5-Hour Sessions Plus Weekly Caps
Every subscription meters usage two ways:
- Five-hour session limit. Usage accrues against a rolling session window. Hit the cap and you wait for the session to reset.
- Weekly limits. Separate weekly caps exist on top of sessions. Max plans have two: one that applies across all models and a second model-specific limit, so you cannot drain the whole weekly budget on the most expensive model.
Both pools are shared between Claude.ai chat and Claude Code. Check consumption in Settings > Usage on claude.ai, or with /usage and /status inside Claude Code. API key usage is a separate billing universe and never touches these caps.
Max 5x vs Max 20x
Anthropic defines the Max tiers relative to Pro: Max 5x ($100/month) provides "5 times more usage per session than the Pro plan", Max 20x ($200/month) provides 20 times more.
| Max 5x | Max 20x | |
|---|---|---|
| Price | $100/month | $200/month |
| Usage per session vs Pro | 5x | 20x |
| Weekly limits | All-models cap + model-specific cap | All-models cap + model-specific cap |
| Agent SDK credit (from Jun 15, 2026) | $100/month | $200/month |
| Daily cost equivalent | $3.33/day | $6.67/day |
Which one you need depends on whether the binding constraint is the session cap or the weekly cap. If Pro stops you mid-afternoon, Max 5x removes that ceiling for most single-developer workloads. Max 20x exists for people running Claude Code most of the working day or driving multiple parallel sessions and subagents.
Team and Enterprise Pricing
Team comes in two seat types. Both include Claude Code; premium seats add Claude Cowork and higher limits.
| Standard Seat | Premium Seat | |
|---|---|---|
| Annual billing | $20/seat | $100/seat |
| Monthly billing | $25/seat | $125/seat |
| Claude Code | Included | Included |
| Claude Cowork | No | Included |
| Agent SDK credit (from Jun 15, 2026) | $20/seat/month | $100/seat/month |
A 10-person team on premium seats runs $1,000/month annual or $1,250 monthly. Enterprise drops the published price list entirely: you negotiate a seat price and usage beyond it bills at standard API rates. Full seat-mix analysis in our Claude Code team pricing guide.
API Pricing Per Token
Skip the subscription, create a Console account, and Claude Code bills per token. No session caps, no weekly caps, spend limits set in the Console. Current first-party rates:
| Model | Input | Output | Cache Read | Batch (in/out) |
|---|---|---|---|---|
| Haiku 4.5 | $1.00 | $5.00 | $0.10 | $0.50 / $2.50 |
| Sonnet 4.6 | $3.00 | $15.00 | $0.30 | $1.50 / $7.50 |
| Opus 4.6 | $5.00 | $25.00 | $0.50 | $2.50 / $12.50 |
Three details third-party pricing pages get wrong:
- No long-context surcharge. Sonnet 4.6 and Opus 4.6 include the full 1M-token context window at standard pricing. A 900k-token request bills at the same per-token rate as a 9k one. The old >200K premium tier is gone.
- Opus 4.5, 4.7, and 4.8 cost the same as 4.6: $5/$25 per MTok.
- Batch is 50% off both directions and stacks with caching, but it is async only, so it fits CI jobs and bulk refactors, not interactive Claude Code sessions.
Full rate tables including cache writes and deprecated models in our Anthropic API pricing breakdown.
Prompt Caching: What Claude Code Tokens Actually Cost
Claude Code resends your system prompt, CLAUDE.md, and conversation history on every turn. Without caching that would be ruinous. With caching, a cache read costs 0.1x base input: $0.30/MTok on Sonnet, $0.50/MTok on Opus. A 5-minute cache write costs 1.25x base ($3.75 Sonnet, $6.25 Opus), so caching pays for itself after a single hit.
Worked example on Sonnet 4.6: a turn that sends 100k input tokens where 90k are cache reads costs 10,000 x $3 plus 90,000 x $0.30 per MTok, or $0.057, versus $0.30 fully uncached. That is an 81% cut on input, which is why long Claude Code sessions cost far less than naive input-tokens-times-rate math suggests. Claude Code manages cache breakpoints automatically; you do not configure anything.
Subscription vs API: The Breakeven Math
Divide the plan price by 30 and compare against your daily API burn:
| Plan | Monthly | Per day | API beats it when |
|---|---|---|---|
| Pro | $20 | $0.67 | You spend under ~$0.67/day in tokens |
| Max 5x | $100 | $3.33 | You spend under ~$3.33/day |
| Max 20x | $200 | $6.67 | You spend under ~$6.67/day |
To estimate your daily burn, run a few representative days on an API key and read the totals from the Console (or watch /cost in Claude Code). With caching active, output tokens usually dominate: at Sonnet's $15/MTok, a day producing 200k output tokens is $3 in output alone before input. Anyone sustaining that daily is past the Max 5x breakeven. Bursty users, one heavy week a month, stay cheaper on API.
The hybrid setup
Common pattern: a Pro or Max subscription for interactive work, plus an API key for CI and batch jobs where the 50% Batch discount applies. After June 15, 2026 the subscription itself covers some automation through the Agent SDK credit below, which shrinks the API-key side of the split.
The June 2026 Change: Agent SDK Credits Split Interactive from Autonomous
Starting June 15, 2026, Anthropic separates human-in-the-loop usage from autonomous usage on subscription plans. Interactive Claude Code sessions keep using the session and weekly limits. Non-interactive usage, meaning Agent SDK projects, claude -p headless runs, the GitHub Actions integration, and third-party apps authenticating with your subscription, draws from a new monthly Agent SDK credit:
| Plan | Agent SDK credit |
|---|---|
| Pro | $20/month |
| Max 5x | $100/month |
| Max 20x | $200/month |
| Team standard seat | $20/seat/month |
| Team premium seat | $100/seat/month |
| Enterprise premium seat | $200/seat/month |
When the credit runs out, usage flows to usage credits at standard API rates if you have enabled them; otherwise non-interactive requests stop until the next cycle. Unused credit does not roll over. If you run nightly agents or CI review bots off a Max plan today, this is the line item to model before June 15.
Claude Code vs Cursor vs GitHub Copilot: Pricing
Entry prices as of June 2026, checked against each vendor's pricing page:
| Tier | Claude Code | Cursor | GitHub Copilot |
|---|---|---|---|
| Free | No Claude Code on free plan | Hobby (limited) | $0, 2,000 completions/mo |
| Entry | $20/mo Pro ($17 annual) | $20/mo Pro | $10/mo Pro |
| Mid | $100/mo Max 5x | $39/mo Pro+ (Copilot) | $39/mo Pro+ |
| Top | $200/mo Max 20x | Ultra tier (see cursor.com) | $100/mo Max |
| Teams | $20-25 std / $100-125 premium per seat | $40/user/mo | Via sales |
Two caveats. GitHub paused new signups for paid individual Copilot plans as of June 2026, so the $10 entry point is not currently purchasable. And raw plan price is the wrong comparison axis anyway: Claude Code is a terminal agent that edits and runs code autonomously, Cursor is an IDE, Copilot is completions plus chat. Capability comparisons live in Claude Code vs Cursor and Claude Code vs Copilot; cross-tool cost modeling in AI coding costs.
How to Reduce Your Claude Code Bill
Whether you are stretching a Pro session cap or an API budget, the lever is the same: fewer tokens per task.
Write a CLAUDE.md under 200 lines
Anthropic's guidance: keep CLAUDE.md under 200 lines with verifiable instructions. It loads every turn, so every line is recurring input cost, but a good one eliminates exploratory file reads that cost far more.
Match the model to the task
Sonnet 4.6 is $3/$15 per MTok vs Opus 4.6 at $5/$25. Switch with /model. Reserve Opus for multi-file reasoning; Haiku at $1/$5 handles mechanical edits.
Start fresh sessions per task
Context resends every turn. Cache reads soften this to 0.1x input, but cache writes and growing history still compound. Discrete tasks in discrete sessions beat one marathon thread.
Use subagents for exploration
A subagent reads 30 files in its own context and returns a summary; only the summary lands in your main session. Parent context stays small, and after June 15, 2026 headless agent runs draw from the separate Agent SDK credit.
Offload edits to a fast-apply model
Morph's morph-v3-fast applies edits at ~10,500 tok/s, so the frontier model emits a short edit snippet instead of rewriting whole files at $15-25/MTok output.
Offload search to WarpGrep
WarpGrep returns the relevant files from a single semantic query instead of a chain of grep-and-read turns. Free for 100k requests, then $1 per 1M requests on Pro.
FAQ
How much does Claude Code cost?
Claude Code is included in Pro ($17/month annual, $20 monthly), Max 5x ($100/month), Max 20x ($200/month), Team ($20-25/seat standard, $100-125/seat premium), and Enterprise. With an API key you pay per token: Sonnet 4.6 at $3/$15 per million input/output tokens, Opus 4.6 at $5/$25, Haiku 4.5 at $1/$5.
Is Claude Code free?
No. The free Claude.ai plan does not include Claude Code access. You need Pro, Max, Team, Enterprise, or a Console (API) account. The cheapest way in is Pro at $17/month billed annually.
What is the difference between Pro and Max?
Price and headroom. Max 5x ($100/month) gives 5 times more usage per session than Pro; Max 20x ($200/month) gives 20 times more, plus two weekly caps (all-models and model-specific). Anthropic no longer publishes exact message or hour counts. If you hit the Pro session cap most working days, Max 5x is the upgrade.
Do Claude Code and Claude chat share usage limits?
Yes. Pro and Max usage is shared across Claude chat and Claude Code under the same five-hour session limit and weekly caps. Track it with /usage and /status. API usage is billed separately and does not affect subscription quotas.
Should I use a subscription or API pricing?
Max 5x costs $3.33/day, Max 20x $6.67/day. If your measured API spend exceeds those figures most days, subscribe. If usage is bursty or under roughly $1/day, pay-as-you-go API wins. Many developers run both: subscription for interactive sessions, API key for batch jobs at the 50% Batch discount.
Is Claude Code cheaper than Cursor?
At entry, both Pro plans cost $20/month (Claude Pro drops to $17 annual). For teams, Cursor charges $40/user/month while Claude Team standard seats are $20-25 and premium seats $100-125. The tools do different jobs, so compare capability per dollar in our Claude Code vs Cursor comparison.
Does Claude Code charge extra for long context?
No. Sonnet 4.6 and Opus 4.6 include the full 1M-token context window at standard per-token pricing. There is no surcharge tier above 200K tokens.
Related Guides
Cut Claude Code token costs with Morph
morph-v3-fast applies edits at ~10,500 tok/s so your frontier model stops rewriting whole files at $15-25/MTok output. WarpGrep finds the right code in one semantic search: free for 100k requests, $1 per 1M on Pro.