Claude Code Pricing (October 2026): Which Plans Include It

Short answer: Claude Code isn't free. It's included with every paid Claude plan: Pro at $20/month ($200/year, about $17/month), Max 5x at $100/month, Max 20x at $200/month, Team ($25 or $125 per seat monthly, $20 or $100 billed annually) and Enterprise ($20/seat/month billed annually, plus usage at API rates). You can also skip subscriptions and pay per token through a Claude Console account. Prices are as of October 17, 2026 and exclude tax.

Every way to pay for Claude Code

The Claude Code product page states the plan positioning above. Claude Code works in your terminal, in supported IDEs, in the Claude desktop app and on the web. Your subscription covers all of them, and they share one usage pool. For setup, see how to install Claude Code.

Subscription vs API billing

Claude Code has two different billing models, and you choose between them by how you sign in.

• Sign in with a Claude subscription (Pro, Max, Team, Enterprise): you pay a flat fee, and Claude Code uses your plan's allowance. The dollar figure in the `/usage` Session block is an estimate at list price. Anthropic's docs say it isn't relevant for billing on Pro and Max.
• Sign in with a Claude Console account: every token is billed at standard API rates from your prepaid balance. On first sign-in, a workspace named 'Claude Code' is created automatically for cost tracking, and you can set spend limits on it.
• Cloud providers (Bedrock, Google Cloud, Microsoft Foundry): billed per token to your cloud account, with spend controls in that provider's billing console.

Max and Team plans include monthly Claude API credits ($100 or $200 on Max, up to $500 pooled on Team), but they don't pay for interactive Claude Code. They don't cover the GitHub Action, IDE extensions, the desktop app or extra usage either. The only Claude Code use they cover is `claude -p` run with an API key from the linked Console organization.

How Claude Code usage limits work

• Shared pool: Claude Code and Claude chat on web, desktop and mobile draw from the same allowance.
• Five-hour window: every plan has a session limit that resets on a rolling five-hour window.
• Weekly limit: paid plans also have a weekly limit across all models, which resets at a fixed time assigned to your account.
• Model-specific limits: Max plans and Team Premium seats get Fable at 50% of weekly limits. Pro can use Fable only through usage credits. If you see 'You've hit your Opus limit', switching to another model family with `/model` keeps you working. A session or weekly limit applies to all models.
• Wrap-up allowance: the Help Center says that when you hit the five-hour limit mid-response, Claude Code may finish the step in progress. This happens once per week on Pro and each time on Max. It's rolling out gradually.
• No published message counts: Anthropic doesn't publish how many prompts or hours each plan allows. The 5x and 20x multiples are the only official comparison.

When you hit a limit, run `/usage-credits` to turn on usage credits. These are prepaid funds billed at standard API rates, with a monthly spending cap you choose. Pro and Max subscribers manage them under Settings > Usage on claude.ai. On Team and Enterprise, members without billing access send a request to their admins. Since v2.1.234, Claude Code can also wait for the reset and continue the interrupted task automatically.

What Claude Code actually costs on the API

Anthropic's cost documentation says that across enterprise deployments, the average is about $13 per developer per active day and $150-250 per developer per month, and 90% of users stay below $30 per active day. Costs vary widely with the model, codebase size and habits such as running several instances or automation.

Coding sessions are dominated by cached context, because Claude Code re-sends the conversation and file contents on every turn. Cache reads are much cheaper than fresh input, which is why caching matters so much. Current per-million-token prices that apply to Claude Code (full table) are:

Worked cost examples

1. One API-billed session on Sonnet 5.5 vs Opus 5.5

Assume a two-hour feature session that reads 2,000,000 cached tokens, writes 200,000 tokens to the 5-minute cache, sends 50,000 uncached input tokens and generates 60,000 output tokens (thinking tokens count as output). Sonnet 5.5: (2 x $0.10) + (0.2 x $2.50) + (0.05 x $2) + (0.06 x $10) = $0.20 + $0.50 + $0.10 + $0.60 = $1.40. Opus 5.5: (2 x $0.20) + (0.2 x $5) + (0.05 x $4) + (0.06 x $20) = $0.40 + $1.00 + $0.20 + $1.20 = $2.80. These volumes are illustrative. Use `/usage` to see your real token mix.

2. Subscription vs API for one developer

At Anthropic's enterprise average of $13 per active day, 20 working days cost 20 x $13 = $260/month on the API. That's more than Max 20x ($200) and 13x Pro ($20). A developer who uses Claude Code only on 4 days a month, at the same average, would spend 4 x $13 = $52, which is less than Max 5x. The break-even depends on how many days you use it, not just how intensively.

3. Pro monthly vs annual

12 x $20 = $240 a year on monthly billing, versus $200 billed annually. That saves $40 (about 17%), in exchange for paying up front.

4. A 6-person engineering team

With 4 Standard and 2 Premium seats on annual billing, the team pays (4 x $20) + (2 x $100) = $80 + $200 = $280/month, or $3,360/year. Billed monthly, the same seats cost (4 x $25) + (2 x $125) = $100 + $250 = $350/month. Compare that with six API users at the documented $150-250 average: 6 x $150 = $900 to 6 x $250 = $1,500 a month. Seats cap your bill but also cap usage, while the API scales with use.

5. A burst of fast mode

Fast mode on Opus 5.5 costs $8 in / $40 out per million tokens. On a subscription, it's paid through usage credits. A burst with 300,000 uncached input tokens and 100,000 output tokens costs (0.3 x $8) + (0.1 x $40) = $2.40 + $4.00 = $6.40, before any cache reads. Turn it on for latency-critical work, not by default.

Which option should you pick?

• Trying Claude Code a few times a week: Pro, monthly at first. Switch to annual once you're sure.
• Hitting Pro limits most days: Max 5x. Move to 20x only if you regularly exhaust 5x.
• Irregular or automated use (CI jobs, scripts): a Console account billed per token, with a workspace spend limit.
• A team that needs SSO and central billing: Team, with Premium seats only for the heaviest users.
• Large organization: Enterprise. Use the Claude Enterprise consumption guide in the Help Center to budget per-user spend.
• Students: there's no individual Claude Code student plan. See Claude student discount.

How to spend less on Claude Code

1. Run `/clear` between unrelated tasks. Stale context is re-sent on every message.
2. Use Sonnet for most work and switch to Opus with `/model` only for hard architectural problems. Set `model: haiku` for simple subagents.
3. Lower the effort level with `/effort`. Thinking tokens are billed as output, and Opus 5.5, Sonnet 5.5, Haiku 5.5 and Fable always use extended thinking.
4. Use plan mode for complex tasks, so wrong directions get caught before Claude writes code.
5. Prefer CLI tools such as `gh` over MCP servers where possible, and disable unused servers with `/mcp`.
6. Move specialized instructions out of CLAUDE.md into skills. Anthropic suggests keeping CLAUDE.md under 200 lines.
7. Use hooks to filter long test or log output before Claude reads it.
8. Be careful with agent teams. Anthropic says teammates in plan mode use about 7x more tokens than a standard session.
9. Avoid long idle gaps mid-task. On a subscription, the prompt cache lasts an hour, and the first message after it expires reprocesses your whole context.

Team and Enterprise cost controls

On Team and Enterprise, admins see estimated spend per user and per model in the org analytics spend report, which exports to CSV. They set usage-credit spend limits at the organization, group or member level under Organization settings > Usage. Enterprise also has the Enterprise Analytics API for per-user cost reports. On the Console, workspace spend limits and the Claude Code Analytics API do the same job. For sizing API rate limits, Anthropic suggests 200K-300K tokens per minute per user for 1-5 users, falling to 10K-15K for organizations of 500+ users, because fewer people use it at the same time. An administrator can also set a `modelPricing` managed setting so Claude Code reports costs at contracted rates instead of list price.

Claude Code vs other coding agents

Details are on our OpenAI Codex pricing, Cursor pricing and GitHub Copilot pricing pages. To compare all Claude plans, including chat-only features, see Claude pricing. If you maintain a popular open-source project, Claude for Open Source offers six months of Max 20x to qualifying maintainers.