Anthropic quietly removed Claude Code from its $20/month Pro plan on April 21, 2026. The pricing page now shows an “X” next to Claude Code for Pro subscribers. Claude Code access now starts at the Max plan ($100/month).
The company says it’s an “A/B test on ~2% of new prosumer signups” and that existing Pro and Max subscribers aren’t affected. But the public pricing page already reflects the change for everyone.
Here’s what you need to know.
What changed
Before April 21:
- Pro ($20/mo): Included Claude Code
- Max ($100/mo): Included Claude Code with higher limits
After April 21:
- Pro ($20/mo): Claude Code removed (marked with X on pricing page)
- Max 5x ($100/mo): Now the cheapest plan with Claude Code
- Max 20x ($200/mo): Higher limits
The Claude Code documentation was updated to reference only the Max plan, removing previous mentions of Pro access.
Who’s affected
Not affected (for now):
- Existing Pro subscribers keep Claude Code access
- Existing Max subscribers keep Claude Code access
- API users (pay-per-token) are unaffected
Affected:
- New Pro signups may not get Claude Code access
- Anthropic says ~2% of new signups are in the test, but the pricing page change is visible to everyone
Why Anthropic did this
Anthropic’s head of growth Amol Avasare explained on X:
“When we launched Max a year ago, it didn’t include Claude Code, Cowork didn’t exist, and agents that run for hours weren’t a thing. Engagement per subscriber is way up. Our current plans weren’t built for this.”
The core problem: Pro subscribers at $20/month are consuming far more compute than the subscription covers, sometimes 10x or more in token value. Claude Opus 4.7, Cowork, and long-running agents made Claude Code dramatically more useful, and dramatically more expensive for Anthropic to serve.
This follows a pattern across the industry. GitHub also paused new Copilot signups the same week, citing similar compute cost pressures.
Your alternatives
If you’re a new user who can’t get Claude Code on Pro, or if you want to prepare for potential future changes:
Stay with Anthropic
- Upgrade to Max ($100/mo): Guaranteed Claude Code access with higher limits
- Use the API directly: Pay per token ($15/$75 per M tokens for Opus 4.7). More expensive per-token but no subscription limits. Works with Claude Code via API key.
Switch to cheaper alternatives
- Kimi K2.6: 80.2% SWE-Bench, $0.60/$3.00 per M tokens. Open-source. Works with Kimi CLI.
- Codex CLI: OpenAI’s terminal agent. Included with ChatGPT Plus ($20/mo).
- Gemini CLI: Free tier with generous limits. Google’s terminal agent.
- Aider: Open-source, works with any model including cheap Chinese models.
Use Claude Code with cheaper models
You can run Claude Code with non-Anthropic models through OpenRouter. Use Kimi K2.6, Qwen 3.6 Plus, or GLM 5.1 as the backend model at a fraction of the cost.
What to expect next
Anthropic promised: “When we do land on something, if it affects existing subscribers you’ll get plenty of notice before anything changes.”
But the direction is clear. Flat-rate unlimited AI coding subscriptions are unsustainable at current usage levels. Expect either:
- Higher prices for Claude Code access
- Token-based billing (similar to what GitHub is exploring)
- Stricter usage caps during peak hours
For a full comparison of AI coding tools and their pricing, see our AI coding tools pricing guide.
FAQ
Is Claude Code still available on the Pro plan?
For existing Pro subscribers, yes. For new signups, Anthropic is testing removal for ~2% of users. The pricing page currently shows Claude Code as not included in Pro.
What’s the cheapest way to use Claude Code now?
If you’re an existing Pro subscriber ($20/mo), you still have access. For new users, the cheapest option is Max 5x at $100/mo. Alternatively, use the API with your own key (pay per token).
Are there free alternatives to Claude Code?
Yes. Gemini CLI offers a generous free tier. Aider is open-source and works with any model. Kimi CLI is free with a Kimi account.
Will existing Pro subscribers lose Claude Code?
Anthropic says existing subscribers are not affected by the current test. But the long-term direction suggests pricing changes are coming.