Anthropic Pulls Claude Code From $20 Pro Plan in 2% Test

Anthropic briefly removed Claude Code from its $20 Pro plan Tuesday, then reversed within hours after developer backlash. Head of Growth Amol Avasare called it a 2% test, though pricing pages and support docs changed site-wide. OpenAI kept Codex free. GitHub paused Copilot signups a day earlier.

Anthropic briefly removed Claude Code from its $20-per-month Pro plan on Tuesday afternoon, flipping the feature to unavailable across pricing pages and support documents before reversing the change within hours. Head of Growth Amol Avasare said the update was a test affecting roughly 2 percent of new prosumer signups, though the grid and documentation changes rolled out across the public site. The reversal followed hours of developer backlash that exposed how little slack sits inside Anthropic's flat-rate subscription economics.

The pricing page swap was first flagged by Ed Zitron on Bluesky. By Tuesday evening, the Pro column on claude.com/pricing showed a red X next to Claude Code instead of the previous checkmark. The Anthropic support article titled "Using Claude Code with your Pro or Max plan" was edited to "Using Claude Code with your Max plan." One word deleted. One access tier gone.

Only the $100-per-month Max 5x and $200-per-month Max 20x tiers retained access on the grid. Avasare's 2 percent framing ran into its own contradiction. A gray-scale experiment targets a slice of new users with a server-side flag. Anthropic pushed the change site-wide instead. Simon Willison, who teaches journalists how to use Claude Code, said he didn't buy the framing: "Everyone I've talked to is seeing the new pricing grid and the Internet Archive has already snapped a copy."

Hours later, Anthropic reverted both the landing page and the documentation. Avasare, on X: "Was a mistake that the logged-out landing page and docs were updated for this test."

Per Avasare, the math is plain. Claude Code usage "took off after Opus 4," Claude Cowork did not exist when Max launched a year ago, and "agents that run for hours weren't a thing." Engagement per subscriber has climbed, and in his words, "our current plans weren't built for this." The Register reported that Anthropic's subscription tiers charge far less than the book value of tokens consumed, sometimes by a factor of ten.

Implicator has traced this math before. A $20 Pro subscriber running Opus can burn close to $60 in inference cost. A $200 Max user runs closer to $570. The Pro price point never paid for heavy agentic workloads, and Anthropic has been pulling back on adjacent giveaways, including the April 4 decision to stop letting Claude subscriptions cover third-party tools like OpenClaw.

Willison captured the damage in plain terms. His trust in Anthropic's pricing transparency has been shaken. He said he would not teach a NICAR data-journalism class built on a $100-per-month subscription. A GitHub issue titled "Breaking Change: Claude Code CLI Removed from Pro Plan Without Notice" went up within hours. A Hacker News thread drew more than 400 comments, with several users reporting immediate cancellations.

The Register flagged the structural risk. Enterprise buyers do not like whipsaw pricing. Individual Pro customers at $20 may not be strategic accounts. The optics reach upmarket anyway.

OpenAI did not miss the window. Thibault Sottiaux, the engineering lead for Codex, posted that the tool will stay in the free and $20-per-month Plus tiers: "We have the compute and efficient models to support it. For important changes, we will engage with the community well ahead of making them." Sam Altman chimed in with "ok boomer."

Google's Gemini CLI stays free. Cursor holds at $20. Chinese models like Kimi, GLM, and Minimax run at a fraction of the cost. If you own decent hardware, open-source code models like Qwen3-Coder and DeepSeek V3 now handle coding jobs locally through Ollama.

And the timing matters. GitHub paused Copilot signups the day before. Both moves landed inside 48 hours, pointing at the same compute wall.

Anthropic now carries what Willison called a "trust bonfire" into its next pricing decision. The company has promised that any change affecting existing subscribers will arrive with notice. Avasare said that message would not come from "a screenshot on X or Reddit." He delivered it in a screenshot on X.

The test is still running, just invisible to the outside world. The pricing grid shows a checkmark again. The receipts do not.