Editorial illustration for Update: Usage Limits Draining Faster Linked to Two Unrelated Experiments
Claude Usage Limits Drain Faster: Community Mystery
Update: Usage Limits Draining Faster Linked to Two Unrelated Experiments
Bugs are supposed to be dumb. This one was clever, which is infinitely worse. It started with user reports: usage limits were draining faster than expected.
The data proved them right. But pinning down the cause was a slog through the dark corners of the codebase.
We believe this is what drove the separate reports of usage limits draining faster than expected. Two unrelated experiments made it challenging for us to reproduce the issue at first: an internal-only server-side experiment related to message queuing; and an orthogonal change in how we display thinking suppressed this bug in most CLI sessions, so we didn't catch it even when testing external builds. This bug was at the intersection of Claude Code's context management, the Anthropic API, and extended thinking. The changes it introduced made it past multiple human and automated code reviews, as well as unit tests, end-to-end tests, automated verification, and dogfooding.
Two harmless experiments, fine in isolation, created a failure that slipped past every checkpoint. Human reviewers missed it. Automated checks missed it.
Unit tests, end-to-end tests, and internal builds all passed. Complex systems fail at the seams between things, not within them. The draining bug is now fixed.
Anthropic's review process now explicitly flags intersecting changes across different stack layers. The lesson is simple, but expensive: a bug can live in the quiet space between two perfectly good ideas.
Related Reading
Open Source Trump cracks down on Anthropic after Amazon tip; staff largely foreign
Open Source SDOF Adds Two Defensive Layers via Intent Router and StateAwareDisp
Open Source D&B rebuilds 642 million‑business database after AI agents hit limits
From Archives Agent Improvement Loop Starts with Trace, Enabling Deterministic, Low‑Cost Validation
From Archives Anthropic's Mythos Leak Precedes Bland AI's Norm Voice Agent Builder
Common Questions Answered
What caused the rapid draining of Claude's usage limits?
Two unrelated experiments were simultaneously running that impacted usage tracking and display. One was an internal server-side experiment related to message queuing, while the other was an orthogonal change in how thinking was suppressed in CLI sessions.
When were the usage limit issues resolved?
The team resolved the three problems on April 20 with version v2.1.116. The API was not modified during these fixes, which addressed issues in Claude Code, the Claude Agent SDK, and Claude Cowork.
Why was it initially difficult to reproduce the usage limit drainage problem?
The independent nature of the two experiments made it challenging for engineers to consistently reproduce the behavior in a controlled testing environment. The orthogonal changes meant that the bug was not immediately apparent during standard testing procedures.
Further Reading
- Claude Usage Limits: Why You're Running Out Faster (And the $338...) — Nicholas Rhodes Substack
- Widespread abnormal usage limit drain across all paid tiers — GitHub - Anthropic Claude Code Issues
- Claude Code users hitting usage limits 'way faster than expected' — Hacker News
- Anthropic tightens usage limits for Claude Code without telling users — Hacker News
- [BUG] Claude Code Usage for Max Plan hitting limits extremely fast — GitHub - Anthropic Claude Code Issues
