Claude Code doesn’t have its own price tag anymore. Anthropic folded it into the same subscription that runs Claude’s chat apps, so what you pay depends on which Claude plan you’re already on (or need to upgrade to), not on a separate Claude Code line item.
The Free plan gets you Claude’s web, iOS, Android, and desktop apps, but not Claude Code. To run the CLI agent you need at least Pro, which is $20/mo billed monthly or $17/mo if you pay $200 upfront for a year. Pro includes “more usage” than Free (Anthropic doesn’t publish a hard number, deferring to its usage-limit best-practices page) plus access to Opus, Sonnet, and Haiku models and Anthropic’s other Pro-tier products like Cowork. For most people doing occasional agentic coding sessions between other work, Pro is where you start.
If you’re running Claude Code most of the day, Pro’s usage ceiling shows up fast. That’s what Max exists for. Max comes in two tiers that differ only in how much usage they include: Max 5x is $100/mo for five times Pro’s allowance, and Max 20x is $200/mo for twenty times Pro’s allowance, confirmed directly on Anthropic’s help center. Both tiers add higher per-task output limits and priority access during high-traffic periods, on top of everything Pro includes. Overage on Max is opt-in rather than automatic: you pick a tier and get that tier’s allowance, and when you exhaust it you either wait for the window to reset, upgrade, or switch on usage credits, which bill at standard API rates. Nothing meters you by default.
Claude Code wasn’t always bundled this way, and it’s worth knowing the history if you’re cross-checking older pricing writeups. It launched in February 2025 as a separate research-preview tool billed through raw API usage, the way you’d pay for the API directly. Anthropic later rolled Claude Code usage into the standard Pro and Max subscriptions, so it now behaves like any other feature gated by your plan tier instead of a metered add-on. If you evaluated Claude Code’s pricing model before that change, treat the API-billing structure as retired; everything now runs through the same subscription math as Claude’s chat products.
Team pricing splits by seat type rather than by feature set. A Standard seat is $25/mo billed monthly ($20/mo if your org pays annually) and includes Claude Code, Cowork, SSO, and central billing and administration — the baseline for orgs of 2 to 150 people. A Premium seat costs five times as much, $125/mo monthly or $100/mo annually, and buys exactly one thing over Standard: 5x more usage per seat. Every other feature — SSO, admin controls, billing — is identical between the two seat types, so the only reason to put someone on Premium is that they’d otherwise hit Standard’s usage ceiling running agentic sessions daily. Anthropic lets you mix seat types within one Team account, so you don’t have to upgrade the whole org to cover your heaviest users.
Enterprise drops the fixed per-seat menu for a $20/seat base plus usage billed at API rates, aimed at organizations that need SCIM provisioning, audit logs, a compliance API, custom data retention, and network-level access controls like IP allowlisting. That’s a materially different cost structure from Team: instead of a flat monthly ceiling, your bill scales with actual model usage on top of the seat fee, which only makes sense once procurement and security requirements are the deciding factor rather than raw cost predictability.
For anyone evaluating Claude Code specifically rather than Claude’s chat interface, the practical decision tree is short. Light, occasional use fits inside Pro’s usage pool. Daily agentic coding work is the exact case Anthropic built Max for, and the honest starting point is Max 5x — upgrade to Max 20x only after you’ve actually hit Max 5x’s ceiling repeatedly, since the jump from $100 to $200/mo is a real one. Teams should default new members to Standard seats and reserve Premium for people running Claude Code as their primary coding interface all day. And if your organization’s blocker is compliance paperwork rather than usage volume, Enterprise’s seat-plus-API-rate model is worth a sales conversation before you commit to per-seat Team pricing that may not include the controls you need.