HeadsUpAI

Anthropic resets Claude Code usage limits after subagent orchestration bug

Anthropic reset the 5-hour and weekly rate limits for all Claude Pro and Max subscribers. The move follows a fix for a Claude Code bug where Opus 4.8 triggered excessive parallel tool calls. This error spawned more subagents—specialized AI units for background tasks—than intended, draining quotas prematurely.
Affected Model
Opus 4.8
Affected Plans
Pro and Max
Reset Scope
5-hour and weekly limits
Bug Cause
Excessive parallel subagents
Resolution
Limits reset for all users

This highlights the fragility of the orchestration layer, the software managing AI-to-tool interactions. While models remained stable, the logic governing their autonomous behavior failed. This mirrors a previous usage limit reset where orchestration bugs caused performance slips, proving that agentic reliability depends on the system harness as much as the model.

Quotas are restored immediately for all subscribers. Anthropic confirmed this bug was unrelated to the recently launched dynamic workflows, which allow Claude Code to orchestrate large agent fleets. The fix ensures that parallel execution remains a controlled feature rather than an unintended cost driver.

ClaudeDevs
ClaudeDevs
@ClaudeDevs
X

We've reset 5-hour and weekly rate limits for all users on Pro and Max plans. We fixed an issue that caused some Claude Code sessions to spawn excessive parallel subagents, burning through usage faster than expected.

1.1kretweets20klikes
View on X

Still wondering? A few quick answers below.

A bug in the Claude Code orchestration layer caused the Opus 4.8 model to spawn unintended parallel subagents and tool calls. This error consumed user quotas much faster than expected, leading Anthropic to manually restore the 5-hour and weekly limits for all affected subscribers.

The usage limit reset applies to all users on the Claude Pro and Claude Max plans. These subscribers should see their 5-hour and weekly rate limits restored to their full capacity immediately, correcting for the unintended usage burn caused by the subagent orchestration bug.

No. Anthropic confirmed that the issue causing excessive subagent spawning was unrelated to the recently introduced dynamic workflows feature. While dynamic workflows are designed to orchestrate large fleets of agents intentionally, this bug was an error in how standard Opus 4.8 requests were handled.

Every HeadsUpAI update is written based on its original source and reviewed before it's published. Read our editorial standards →

Share this update