[BUG] Max plan ($200/mo) hitting usage limits unreasonably fast during business hours

Preflight Checklist

• [x] I have searched existing issues and this hasn't been reported yet
• [x] This is a single bug report (please file separate reports for different bugs)
• [x] I am using the latest version of Claude Code

What's Wrong?

What happened
I'm on the Claude Max plan at $200/month. During a single working session today (roughly 1 hour), I hit usage limits while doing legitimate, productive software engineering work. This is a recent change — the same type of session would not have burned through limits this quickly before.

The complaint
Something has changed on the backend that is throttling or limiting usage during business hours (US Eastern, daytime) far more aggressively than before. I suspect an A/B test or capacity management change was rolled out without communicating it to paying users.

If Anthropic is going to rate-limit $200/month subscribers during the hours they actually need to work, that needs to be disclosed _transparently_ — not shipped silently. The value proposition of Max is that it's supposed to be the plan for power users who need sustained, heavy usage. If that's being quietly degraded, people need to know before they commit to paying for it.

Environment
Plan: Claude Max ($200/mo)
Tool: Claude Code (CLI/VSCode extension)
Model: Claude Opus 4.6 (1M context)
Date: 2026-03-26, US Eastern business hours

Please advise on next steps.

What Should Happen?

Claude should not be rate limiting Max 200 plans for this level of work.

Error Messages/Logs

Steps to Reproduce

What I accomplished in that session
In one conversation with Claude Code (Opus), I completed the following across a full-stack TypeScript/React/Node.js platform:

ClickUp project management (via MCP):

Reviewed the full task board, pulled task details, resolved member IDs
Closed 3 completed tasks with detailed comments
Created 4 new follow-up backlog tasks with descriptions
Updated priorities and statuses on multiple tasks
Reviewed and triaged 14 open tasks assigned to the AI contributor

Bug fixes (2):

Fixed a CSS viewport overflow bug in a desk/canvas layout component
Fixed a sidebar flash bug by decoding JWT payload on initial auth hydration
New feature — Share Sheet:

Built a new ShareSheet component (copy link + send in-app notification)
Added a backend API endpoint (POST /notifications/send)
Integrated into RecordHeader and wired it into all 12 detail pages
Both web and API type-check clean

Source control:

Committed, pushed, and created a PR (#16) to main with full summary and test plan

Planning:

Explored 3 additional feature areas (edit slideout, activity controls, duplicate detection) for the next sprint
Reviewed newly assigned tasks with updated descriptions
This is exactly the kind of work the Max plan is supposed to support. None of this was frivolous — it was a focused engineering session that produced a real commit, a real PR, and real project management.

Claude Model

None

Is this a regression?

Yes, this worked in a previous version

Last Working Version

_No response_

Claude Code Version

Claude Code version: 2.1.72

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

VS Code integrated terminal

Additional Information

_No response_