Anthropic Quietly Pulled Claude Code From the $20 Pro Plan

Sometime around April 21, Anthropic updated its pricing page. Claude Code, previously included with the $20/month Pro subscription, now shows an “X” under both the Free and Pro tiers. The only plans with Claude Code access are Max 5x at $100/month and Max 20x at $200/month.

There was no blog post. No email to subscribers. No changelog entry. Developers found out by looking at the pricing page.

Anthropic Says It’s a Test

Amol Avasare, Anthropic’s head of growth, responded to the backlash by clarifying the scope: the change affects roughly 2 percent of new “prosumer” signups. Existing Pro and Max subscribers keep their current access. This is framed as a small experiment, not a permanent policy shift.

Whether a 2 percent test on new signups stays at 2 percent is a different question. The pricing page update was visible to everyone, which is why it spread so fast.

Why This Is Happening

The economics of AI subscriptions have changed dramatically since Anthropic launched its pricing tiers. When the Max plan debuted a year ago, Claude Code didn’t exist yet. Neither did Cowork. Agents that run for hours, consuming thousands of tokens per session, weren’t part of the calculation.

The gap between what subscribers pay and what their usage actually costs has grown. Subscription plans sometimes charge a factor of ten or more below the book value of the tokens consumed. A developer running Claude Code heavily on a $20 Pro plan can burn through compute that would cost hundreds on the API. That math doesn’t work at scale, especially for a company that just raised $30 billion at a $380 billion valuation and needs to show a path to profitability.

The Developer Reaction

The response was swift and mostly negative. Ed Zitron, a vocal AI industry critic, flagged the pricing page discrepancy early. One developer’s tweet about the change hit nearly 900,000 views. Reddit threads accumulated hundreds of comments.

On GitHub, someone filed issue #45236 titled “Breaking Change: Claude Code CLI Removed from Pro Plan Without Notice.” The title captures the core frustration: developers who signed up for Pro specifically because it included Claude Code feel like the terms changed under them. They paid for one thing and got a different thing, with no heads-up.

The word “bait-and-switch” appeared in nearly every thread. Whether that label is fair depends on how you view the “2 percent test” framing. If you’re in the 2 percent, the distinction between a test and a policy change is academic.

The Competitive Angle

The timing handed competitors a talking point. OpenAI launched a new $100 ChatGPT Pro plan the same week, positioned directly against Claude Max. The pitch to frustrated Claude developers writes itself: same price bracket, comparable coding agent access, and a company that isn’t experimenting with taking features away.

The move also strengthened the case for local models. Tools like Ollama and LM Studio let developers run coding agents without depending on a provider’s pricing decisions. The quality gap between cloud models and local alternatives has been narrowing, and every time a cloud provider pulls a feature or raises prices, the calculus shifts a bit further toward local.

Several developers in the Hacker News thread said they’d already started running local model setups alongside their Claude subscriptions as insurance. The Anthropic pricing change gave them a reason to accelerate that transition.

What This Means Going Forward

Anthropic called this a test, and tests can go either way. They might restore Claude Code to Pro, adjust the limits, or introduce a new tier between Pro and Max. Or the test results might confirm what the pricing page already suggests: Claude Code at $20/month isn’t sustainable with current usage patterns.

The broader pattern here isn’t unique to Anthropic. Every AI company is working through the same tension between subscriber growth and unit economics. Flat-rate subscriptions that include agentic tools with high per-session costs are inherently unstable. The question isn’t whether pricing will change. It’s how much notice you get and whether the new terms are reasonable.

For developers relying on Claude Code, the practical move is the same one that applied after the third-party tool lockout in April: don’t build a workflow that depends on a single provider’s pricing staying fixed. Keep an API key handy, know what your usage actually costs, and have a fallback. The developers who got through January’s OAuth changes with minimal disruption were the ones who’d already built with provider portability in mind.