Anthropic Briefly Pulls Claude Code From $20 Pro Plan, Reverses Within Hours

Anthropic quietly removed Claude Code from its $20-a-month Pro subscription on the evening of April 21, 2026, then reversed the change in less than 24 hours after public backlash. Head of Growth Amol Avasare disclosed on X that the company had been "running a small test on ~2% of new prosumer signups," but the silent edit to public pricing pages and support documentation rattled developers already squeezed by tightening rate limits.

Key Highlights

Pricing page edit on April 21–22, 2026: The Claude Code checkmark in the Pro column was replaced with an X, and support docs briefly read "your Max Plan" instead of "your Pro or Max Plan."
Reversal in under 24 hours: Both the landing page and documentation were restored after the change went viral.
Anthropic's framing: A "small test on ~2% of new prosumer signups" with "existing Pro and Max subscribers" unaffected, per Avasare.
Underlying driver: Long-running coding agents now make thousands of API calls per session versus around 50 per day in 2024, breaking older subscription economics.
Competitive fallout: Developers are openly migrating to OpenAI Codex, whose new $100 ChatGPT Pro tier launched on April 9, 2026 with 5x the Codex usage of the Plus plan.

What Happened

Researcher and developer Simon Willison was first to publicly flag the edit. Internet Archive snapshots confirmed that a checkmark next to Claude Code in the Pro column on claude.com/pricing was swapped for an X, and the Claude Code product documentation was rewritten to reference only the Max plan. There was no announcement, no email to subscribers, and no blog post.

Within hours, the change was trending on developer X and was picked up by The Register, The New Stack, and Simon Willison's own blog. Avasare responded on X: "For clarity, we're running a small test on ~2% of new prosumer signups. Existing Pro and Max subscribers aren't affected." He later acknowledged: "was a mistake that the logged-out landing page and docs were updated for this test." Anthropic then restored both the pricing page and the documentation.

Why Anthropic Is Stress-Testing Pricing

Avasare framed the experiment as a response to changing usage patterns. "The way people actually use a Claude subscription has changed fundamentally," he wrote. "We've made small adjustments along the way (weekly caps, tighter limits at peak), but usage has changed a lot and our current plans weren't built for this."

The numbers behind that statement are striking. According to industry reporting around the incident, an average user in 2024 made around 50 model calls per day. A 2026 developer running Claude Code or an agentic Codex workflow can fire off thousands of calls in a single session, with parallel agents and hours-long autonomous loops. The $20 Pro tier, designed for chat usage, was never sized for this kind of consumption.

Developer Backlash and the Codex Migration

Before the reversal, developers vented across X and Reddit. Many reported hitting Claude Code rate limits within 20 minutes on the Max plan during peak hours, with some saying they had already cancelled their subscriptions. Posts shared widely on X described teams keeping both Claude Code and Codex configured to switch mid-session whenever one tool throttled.

The timing matters because OpenAI moved first. On April 9, 2026, OpenAI launched a $100 per month ChatGPT Pro tier explicitly aimed at heavy Codex users, offering roughly 5x the Codex usage of the $20 Plus plan and accommodating higher throughput before throttling. Trending X searches in the days after Anthropic's experiment surfaced the phrase "great migration to Codex," and posts asking developers whether they had bailed on Claude over rate limits drew thousands of replies.

Willison flagged three concerns in his analysis: the abrupt jump from $20 to a $100–$200 Max plan would create real hardship for users globally; the lack of a formal announcement eroded trust; and Anthropic risked surrendering Claude Code's category lead to competitors offering lower-cost access.

Impact on Developers and Teams

For agencies and indie developers in markets where every dollar of subscription cost matters, the incident is a warning. The Pro plan still includes Claude Code today, but Anthropic has signalled that the economics of $20 access for autonomous coding agents are unsustainable as currently structured. Even a 2 percent A/B test on a public pricing page is enough to trigger a confidence shock.

For team leads, the practical takeaway is to budget for tier changes, design workflows that can fall back across providers, and watch the usage caps closely. Several developers in the threads around the incident described porting Claude Code workflows to Codex multi-agent setups in a single sprint as a hedge.

What's Next

Anthropic has not announced a permanent pricing change, but the company is clearly evaluating options. Avasare said Anthropic is "looking at different options to keep delivering a great experience for users," which observers read as a signal that further weekly caps, peak-hour throttling, or new tiers above Max are on the table.

For the broader AI coding market, the episode crystallizes a structural shift: subscription economics built for chat are colliding with agentic workflows, and the next twelve months will reshape how developers buy AI coding capacity. Expect more tiered plans, more usage-based add-ons, and more cross-provider tooling that lets teams switch models when limits hit.

Source: Simon Willison's Weblog, The Register, The New Stack