Claude Code is now included with Pro and Max plans

Running AI agents in your terminal used to come with constant anxiety over runaway API token bills.

Anthropic just removed that friction for individual subscribers. Starting June 4, Claude Code is officially included with Claude Pro and Max plans, letting developers invoke the tool directly from their terminal or IDE without maintaining a separate API pay-as-you-go balance.

Coding in terminals and IDEs, with a hidden billing catch

Claude Code is not limited to bare-metal command lines. The update also covers major code editors, including VS Code, Cursor and other VS Code forks, as well as JetBrains IDEs like IntelliJ and PyCharm.

Developers simply log in with their existing Claude Pro or Max credentials across these environments to start delegating repository tasks.

There is, however, one critical gotcha to watch out for. If your local machine has the ANTHROPIC_API_KEY environment variable set, Claude Code will prioritize that API key over your active subscription. That means your tasks will incur regular per-token API charges instead of consuming your prepaid plan quota.

Anthropic recommends running the logout command, updating Claude Code, and signing in cleanly with your subscription account if you previously used console credentials.

Shared quotas and a safety wrap-up buffer

While this spares developers from per-query billing, it does not mean unlimited usage.

Claude Code draws from the exact same usage pool as your Claude web, desktop, and mobile app interactions. Both Pro and Max accounts remain bound to shared five-hour session caps and rolling weekly limits.

To keep an agent from abruptly breaking mid-task when you hit a cap, Anthropic is rolling out a small wrap-up allowance. If a session limit hits mid-response, Claude Code displays a warning and finishes the current step instead of quitting cold. Pro plans get this allowance up to once per weekly cycle, while Max subscribers can trigger it during each session boundary.

Once caps are reached, developers can wait for limits to reset, buy usage credits, upgrade to higher Max tiers, or manually opt into API billing for urgent coding sprints.

Why it matters

We think folding Claude Code into flat-rate consumer tiers is a calculated offensive against third-party coding tools.

Terminal coding agents have always suffered from an adoption barrier: developers dread leaving an autonomous loop running when every file inspection burns API credits. Giving Pro users a predictable subscription sandbox dramatically lowers the psychological friction to experiment.

For the broader industry, this tightens the screws on AI coding wrappers like Cursor as well as GitHub Copilot. By packaging frontier reasoning and a native agentic workflow under one consumer subscription, Anthropic is turning local development into a standard feature rather than an expensive enterprise add-on.

The catch is the shared quota system. Deep codebase scans eat tokens at breakneck speed, meaning heavy terminal sessions could easily freeze your web chat for the rest of the afternoon. Anthropic will also need to fix the silent API key override, which risks accidentally billing users who thought they were safely running on flat subscription time.