Anthropic finds no regression behind Claude Code usage complaints, and fixes a few token-burn bugs

Boris Cherny of the Claude Code team posted on X on October 28 that the team went through all the session IDs users shared after complaints that Claude Code was using up limits faster. He says they found a few smaller issues that made some sessions burn through tokens faster than they should, and that fixes are rolling out now.

According to Cherny, the team found no single root cause and no regression in Claude Code. He says most users who saw faster usage against their limits had one or two heavy sessions, using xhigh effort, /goal, Workflows, Agent Teams or long SDK jobs. He says these features are fine to use but their cost should be made clearer.

His advice: keep an eye on the effort level, since subagents inherit it by default and the cost can add up fast, and use /usage to see where a session's tokens went. He asks users who still see problems to send /feedback, and says the team will follow up directly with everyone who shared an ID.

This is Anthropic's own account of its investigation; the findings are not independently verified.