How Much Does Claude Code Cost? Plans, Limits and the API Key Trap

“What does Claude Code cost?” The short answer: the program is free, but using the models is not. You need an eligible paid Claude plan or usage-based API billing.

I use Claude Code daily. Here is how the plans and limits worked when checked, and how an unexpected billing configuration caught me out.

Pricing checked November 14, 2026 against Anthropic's pricing page. These are US web prices before applicable taxes. Plans, regional prices and limits can change; check the provider before deciding.

Individual plans at a glance

Team and Enterprise have separate arrangements. Actual usage depends on the model, project and conversation size; the plan figures are not guaranteed numbers of tasks.

Max and Ultra also include monthly API credits of $100 and $200 respectively. According to Anthropic, those credits do not cover interactive Claude Code sessions or extra usage in the Claude apps. Anthropic describes an exception for self-run claude -p with an API key from an appropriately linked Console organization; check its conditions in the source.

How limits work

Claude chat and Claude Code share usage within a plan. Pro and Max have a five-hour window and a weekly limit. Verified Ultra accounts have no five-hour limit but still have weekly limits; model-specific limits can also apply. Your account's usage view shows your own current status.

Once a limit is reached, you can wait for it to reset, use eligible paid extra usage if enabled, or change tools. I often switch: when my Claude allowance is exhausted, Codex can continue. I wrote about the handover in Using Claude Code and Codex together.

The trap: a configured API key

Anthropic says an ANTHROPIC_API_KEY in the environment can take precedence over a Claude subscription, resulting in separate API charges. See its plan and billing guidance.

I experienced a related surprise: Claude Code showed API billing rather than my plan. In my case the problem was an old authentication profile, not a key in the shell. After removing the old profile, I used the plan again.

To check whether this shell has a key without printing its value:

if [ -n "${ANTHROPIC_API_KEY:-}" ]; then
  echo "API key is set"
else
  echo "No API key is set in this shell"
fi

In Claude Code, /status can help you inspect the active account and usage. Check Console billing and any extra-usage settings as well; a key's presence in one shell does not describe every possible environment or authentication method.

Which plan?

My rule of thumb, not a benchmark:

• Occasional small projects: start by considering Pro; a usage-based API may suit infrequent use if spending is controlled.
• Daily work that repeatedly reaches Pro limits: consider Max.
• Regularly hitting Max limits: consider Ultra. Its five-hour window is removed for verified accounts; its weekly limit remains.
• Automated applications: consider a separate API setup and explicit spending controls.

I use Ultra for work across many projects and keep Codex available when a limit gets in the way. Your usage and costs may differ.

What the price does not show

Time spent waiting for a limit or correcting work that should not have been built can matter too. Clear tasks, a concrete definition of done, focused instructions and an appropriate second tool help me keep work moving. They are practices from my experience, not measured savings.

Takeaways

1. Claude Code is included from the paid Pro plan; the free Claude plan does not include it.
2. Chat and code share plan limits. Verified Ultra accounts have no five-hour window, but still have a weekly limit.
3. Monthly API credits and paid extra usage are distinct; read the provider's conditions for each.
4. Check whether an API key is set without displaying it, then inspect /status and billing settings.

Prices and limits change. Check the linked provider pages before making a decision.