Anthropic · Pricing
Claude Code has no price of its own. It ships inside a Claude subscription from $20 a month, inside a Team seat, or on a raw API key billed per token with no allowance at all. There is no free tier. Which route is cheapest depends entirely on how many hours a day you actually run it, so the four are laid out side by side below.
Opens on a coding-agent day rather than a chat day — the inputs are set so the API column lands inside Anthropic's own published band for per-developer spend, printed further down this page. Change them to your volume; every plan re-judges as you type.
API side uses the vendor's published per-token list prices with prompt caching modeled (chat apps resend conversation history every turn — set cache to 0 for a stack without caching). Thinking models bill hidden reasoning tokens at the output rate — that's real spend in both worlds.
Plans cap usage in rolling windows and vendors don't publish those caps in tokens — verdicts compare the price of this volume, not entitlements. The higher the multiple, the more likely a cap cuts in first.
Most pages answer "what does Claude Code cost" with one number. That number is the cheapest subscription that happens to include it, which is the right answer for exactly one kind of buyer. Anthropic sells access four ways, and they are priced in different units — per person per month, per seat, per token, and per whatever your cloud charges. Every figure below is read off Anthropic's own pages on the date shown in the byline.
| Route | Bought as | Price | What it buys |
|---|---|---|---|
| Claude Pro | Subscription | $20/mo $17/mo billed annually | Entry point for Claude Code. The allowance is shared with Claude chat, so terminal work and chat draw from one pool. |
| Claude Max 5x | Subscription | $100/mo | 5x Pro capacity per session. Monthly billing only. |
| Claude Max 20x | Subscription | $200/mo | 20x Pro capacity per session. Monthly billing only. |
| Team · standard seat | Per seat | $25/seat/mo $20/mo billed annually | Pricing page lists the standard seat as including Claude Code and Claude Cowork. |
| Team · premium seat | Per seat | $125/seat/mo $100/mo billed annually | 5x the usage of a standard seat. Premium seats carry two weekly limits: one across all models, one for Sonnet models only. |
| Enterprise | Per seat | Contact sales | No published price. On new and self-serve Enterprise plans Claude Code comes with every seat; older plans split seats into chat-only and Claude Code types. |
| Claude Console API key | API key | At model rates | Billed per token at the model's published rates, with no plan allowance. Prompt caching keeps a one-hour cache lifetime on a subscription; on an API key it is five minutes by default. |
| Cloud providers | Marketplace | Set by the cloud | Billed per token to your cloud account, and spend controls live in the cloud provider's console. The rate is the cloud's, not Anthropic's, so no figure is carried here. |
The free Claude plan does not include Claude Code — its feature list never names it, and Anthropic scopes the tool to Pro, Max, Team and Enterprise. There is no trial tier.
Read off, in order of appearance: claude.com/pricing, 2026-08-18 support.claude.com/choose-a-claude-plan, 2026-08-18 support.claude.com/use-claude-code-with-your-team-or-, 2026-08-18 code.claude.com/costs, 2026-08-18
The entry tier is the honest starting point, and the allowance is shared with Claude chat — if you already use the chat heavily, the terminal draws from the same pool and you will hit the five-hour window sooner than the price suggests.
This is where most full-time users land. Anthropic's own per-developer figure for enterprise deployments — the one printed further down this page, in the vendor's words — crosses the entry tier inside a working week. Read it before assuming yours will be lower.
Agent teams spawn a Claude Code instance per teammate, each with its own context window, and Anthropic puts plan-mode teams at roughly seven times the tokens of a standard session. That pattern, not typing speed, is what exhausts the lower tiers.
Seats carry their own allowance and admin-side spend controls, which individual plans do not. The premium seat exists for the same reason the higher personal tiers do — heavier users on the same team.
No allowance to waste and no ceiling to hit, but no cushion either: you pay for every token, and the prompt cache lifetime drops from an hour to five minutes, which makes long idle gaps expensive. Run your own numbers before choosing this.
The API route is the one people underestimate, because it looks like the flexible option and behaves like the expensive one.
On an API key there is no plan allowance: Claude Code bills every token at the model's published rate, which lives in our Anthropic rate table and moves when Anthropic reprices. That is the whole appeal — nothing is capped, nothing resets, and a quiet month costs almost nothing.
The catch is caching. Claude Code re-sends the whole conversation with every request and relies on the prompt cache to make that affordable. On a subscription the cache lives an hour; on an API key it is five minutes by default. A coffee break is long enough to miss it, and the next message reprocesses your entire context at full rate. That single difference is why per-token billing so often lands above the subscription it was meant to undercut.
Two other things travel with this route: spend controls move to the Claude Console rather than your account settings, and the usage figure Claude Code prints in the session is computed locally from list rates, so it will not match a contracted or promotional bill.
These three figures are Anthropic's, not ours. They come from the vendor's own cost documentation, describe enterprise deployments rather than individuals, and are the numbers every SaaS blog in this search result quotes without saying whose they are. We cite them because they are the only per-developer data anyone has published — and we attribute them, because adopting someone's marketing arithmetic as a finding is how a price desk loses its usefulness.
The part vendors bury and comparison sites skip.
Cancelling takes effect at the end of the current billing period, and the plan keeps working until then — you are not cut off the moment you click. To avoid the next charge, Anthropic asks you to cancel at least 24 hours before the renewal date; inside that window the next period bills and then ends.
Refunds are the harder part. Anthropic's policy is that, except where its consumer terms or the law say otherwise, all payments are non-refundable. Customers in the EEA and the UK are the documented exception and may request a refund within 14 days of purchase, prorated by usage. One trap worth knowing: requesting a refund does not cancel anything. If you want the charges to stop, you have to cancel separately.
This is the argument for monthly billing over the annual commitment on the entry tier while you are still deciding. The annual price is lower per month, and it is also twelve months of a non-refundable payment on a tool whose usage limits you have not yet tested against your own working day.
What people search next.