Anthropic tests Claude Code removal
- Anthropic briefly removed Claude Code from its $20 Pro plan, then said the change affected only a small subset of users as a test. - The move sparked discussion about unbundling high‑intensity features like coding from general subscriptions. - The episode signals vendors are experimenting with pricing segmentation for agentic and compute‑heavy workloads (theregister.com).
Anthropic briefly tested removing Claude Code from its $20-a-month Pro plan, then said the change hit only a small slice of new users. (theregister.com) The company’s public pricing page now lists Pro at $20 monthly, or $17 a month on an annual plan, and Max starting at $100 a month. The same page shows Claude Code as a feature row without a checkmark under Pro, while Max includes higher usage limits and early access to advanced features. (claude.com) Anthropic’s product page describes Claude Code as an “agentic” coding system that can read a codebase, edit files across projects, run tests, and use command-line tools. In practice, it is closer to a software worker than a chat box, which makes it more compute-intensive than ordinary question-and-answer use. (anthropic.com) Ars Technica reported that developers noticed the change on April 21, and Anthropic later said it was testing the removal with a small subset of users rather than making a full plan change. The Register reported the test covered about 2 percent of new signups. (arstechnica.com, theregister.com) The episode lands as AI companies are separating everyday chat from heavier-duty tools that consume more model time and more infrastructure. Anthropic’s Max plan promises 5x or 20x more usage than Pro, a pricing structure built around customers who push the system much harder. (claude.com, claude.com) Anthropic has already used that logic in business plans. In August 2025, it said Team and Enterprise customers could buy premium seats that bundled Claude Code with more usage, while standard seats stayed on the regular Claude app. (anthropic.com) That split reflects how these tools work. A coding agent that scans repositories, runs tests, retries failures, and handles long tasks in the background can burn through far more tokens and compute than a user drafting emails or summarizing documents. (anthropic.com, anthropic.com) Anthropic has not announced a permanent end to Claude Code in Pro, and its current pricing page still says plans and prices can change at Anthropic’s discretion. For now, the company has shown that even consumer AI subscriptions are being carved up by workload, not just by brand or model access. (claude.com, theregister.com)