Recent complaints from Claude users about reaching their access caps sooner than expected are valid, as the company has verified the changes.
Anthropic disclosed that it has been altering the five-hour restrictions for its Free, Pro, and Max tiers specifically during busy periods from 5 a.m. to 11 a.m. Pacific Time on weekdays, without affecting the total weekly allowances.
This information emerged from a Reddit announcement by an Anthropic official, which the publication verified through direct contact with the company.
Although the exact start date for these modifications wasn't detailed in the announcement, reports indicate the updated restrictions began on Monday.
Over the last week, numerous Claude subscribers expressed frustration over accelerated limit encounters, with many theorizing a quiet cutback in their hourly quotas—a suspicion that proved accurate.
The company explained the changes as a measure to handle increasing demand for the AI service.
Anthropic noted achieving several improvements in efficiency to counterbalance the adjustments, though approximately 7% of users, especially those on Pro plans, may now encounter session caps that previously wouldn't apply. For those running high-token background tasks, the suggestion is to schedule them during less busy times to extend available usage.
The firm recognized the inconvenience caused by the tweaks and stated it remains committed to enhancing capacity in a sustainable manner.
This development in curbing peak-hour access coincides with heightened attention on Claude amid its recent dispute with the U.S. Defense Department, where officials attempted to classify the company as a potential 'supply chain risk' due to its refusal of a defense-related agreement. A court recently halted the department's effort to impose that designation.
Anthropic's decision to revise its five-hour caps highlights a wider challenge in the AI sector: the management of user access for those on subscription-based models with fixed fees.
Historically, subscribers to premium tiers—priced between $10 and $250 monthly across various providers—seldom encountered usage barriers when simply engaging in conversational interactions via online interfaces.
However, the emergence of advanced capabilities like agent-driven tools for creative coding and 'computer use' features has led to significantly higher token consumption among paying users, putting pressure on major AI firms to address surging needs.
This issue is intensified by the introduction of Claude's expansive one-million-token context capability earlier this month.