Claude Code Pricing: What Teams Actually Pay in 2026

Claude Code is Anthropic's command-line coding agent. You can pay for it two ways. One is a flat monthly subscription. The other is pay-as-you-go API tokens.

This matters for teams. A subscription gives a predictable per-seat bill. API billing scales with real usage. Most teams start on subscriptions and add API access for automation.

Below we map every tier and the numbers behind them. We also flag the "for teams" trade-offs: cost control, admin power, and data handling. All prices are as of July 2026 and can change, so verify current pricing on Anthropic's page before you buy.

Two ways to pay for Claude Code

Claude Code bills either through a Claude subscription or through Anthropic API tokens. A subscription bundles Claude Code into a flat monthly fee per person. API billing charges you per token of input and output.

Subscriptions suit steady daily coding. The cost is predictable, so budgeting is easy. API billing suits scripts, CI jobs, and heavy automation where usage swings.

Claude Code is included on paid Claude plans (Pro and above), per Anthropic's pricing page. You do not buy Claude Code as a separate product.

• Subscription: flat monthly fee, predictable, best for daily human coding.
• API tokens: usage-based, best for automation and variable workloads.
• Many teams run both: seats for developers, API keys for pipelines.

Claude subscription tiers (Pro, Max, Team, Enterprise)

Claude subscriptions run from a free tier up to Enterprise, and Claude Code is included from Pro upward. As of July 2026, Anthropic's pricing page lists the tiers below.

Pro costs $20 per month, or $17 per month billed annually. It unlocks Claude Code, more usage than Free, and multiple Claude models.

Max starts at $100 per month. You choose 5x or 20x more usage than Pro, plus higher output limits and priority access during busy periods.

Team is built for groups of 5 to 150 people. A standard seat is $25 per month, or $20 per month billed annually. A premium seat, with more usage, is $125 per month, or $100 annually. Team adds central billing, SSO, and admin controls.

Enterprise is roughly $20 per seat plus API usage costs, with a custom contract. It adds SCIM, audit logs, a compliance API, custom data retention, and a HIPAA-ready option.

• Free: $0, standard limits, no Claude Code.
• Pro: $20/mo ($17/mo annual), includes Claude Code.
• Max: from $100/mo, 5x or 20x Pro usage.
• Team standard seat: $25/mo ($20 annual); premium seat: $125/mo ($100 annual); 5-150 people.
• Enterprise: ~$20/seat plus API usage, custom contract, SSO/SCIM/audit logs.

API token pricing and how usage adds up

Anthropic's API charges per million tokens, split between input and output, and priced by model. Cheaper, faster models cost less per token; the most capable models cost more.

Claude Code can run against your API key instead of a subscription. Your bill then tracks the tokens each session reads and writes.

This model rewards efficient prompting. Large codebases and long agent runs use more tokens. Caching and smaller models can cut the cost of repeat context.

Because token rates change with each model release, we do not quote fixed per-token numbers here. Check Anthropic's API pricing page for the current rate per model before you forecast a budget.

• Billed per million input and output tokens, by model.
• Big repos and long agent runs raise token counts.
• Prompt caching and smaller models reduce repeat cost.

Subscription vs API: which is cheaper for a team?

For steady human coding, subscription seats are usually cheaper and easier to budget than API tokens. The flat fee caps your spend per person.

API billing wins when usage is spiky or automated. A nightly CI agent may cost pennies some days and dollars on others. Paying per seat for that would waste money.

The practical answer for most teams is both. Give each developer a subscription seat for daily work. Use a metered API key for pipelines, bots, and one-off batch jobs.

• Predictable daily coding: subscription seats.
• Spiky or automated workloads: API tokens.
• Set an API spend cap so automation cannot run away with the budget.

Security, IP, and data handling for teams

On Claude business plans, Anthropic states it does not train its default models on your content. That protects your source code and prompts from becoming training data.

Team and Enterprise add the controls a business buyer needs. These include SSO, admin roles, audit logs, and, on Enterprise, custom data retention and a HIPAA-ready offering.

Before rollout, confirm which repos and secrets the agent can reach. An AI coding agent runs commands, so scope its access like you would any powerful tool.

• No default model training on business-plan content (per Anthropic).
• Enterprise adds SCIM, audit logs, compliance API, custom retention, HIPAA-ready.
• Scope repo and secret access before giving the agent keys.

The ROI angle for engineering leaders

The right question is not the seat price. It is whether a seat saves more engineer time than it costs. A $25 seat pays for itself if it saves even 15 minutes of a developer's week.

Start with a small pilot on Pro or Team seats. Measure time saved on real tasks like tests, refactors, and boilerplate. Then decide whether to scale.

For automation, price the API path against the hours it removes. A CI agent that triages failing tests can cost a few dollars a day and save a person hours.

• Compare seat cost to engineer hours saved, not to other tools' sticker price.
• Pilot on a few Team seats, measure, then scale.
• For pipelines, weigh API spend against hours removed.