Anthropic has revised the method for applying time constraints for Claude users as a means to balance demand against the available computational resources. During peak usage periods, the five-hour session limit for Free, Pro, and Max subscribers may now deplete more rapidly, although the weekly allowances remain unchanged.
Image Source: Getillustrations / unsplash.com
Image Source: Getillustrations / unsplash.com
This adjustment was announced by Anthropic technical team member Thariq Shihipar. He estimates that approximately 7% of users will encounter session limits in situations where they previously did not, primarily affecting those on higher-tier plans. Peak hours are defined as the interval between 05:00 and 11:00 Pacific Time. During this timeframe, usage equivalent to five hours might be exhausted more quickly. Outside of this window, the five-hour limit functions under its usual consumption rate.
Anthropic ties hourly usage to token consumption without disclosing the specific token count that equates to the time limit, thus preventing users from proactively managing their token expenditure. Anthropic’s documentation states that dialogue length and complexity, features utilized, and the specific Claude model selected all influence usage volume. Furthermore, different subscription tiers, including Pro, Max, and Team, come with distinct allowance levels.
Users initiating background tasks that involve substantial token usage are advised to shift this workload to off-peak hours. The company clarifies that during periods of lower demand, computational capacity is greater, meaning the total weekly allowance is not reduced; only its distribution across the week is altered. A monitoring dashboard within the Claude interface allows users to track both the consumption of their five-hour session limits and their weekly allowances. Once a limit is exhausted, access is blocked unless the user has opted to pay for additional usage.
Anthropic offers access to its AI services via two main avenues: an API for developers and a subscription model for general Claude users. With the API, costs are calculated based on published rules concerning different token operation types—the units of text processed and generated by the AI model. Subscribers to Claude operate under a different structure: instead of a transparent calculation based on operations, they are subject to internal usage caps, the precise methodology for which the company does not reveal. Current offerings include Free, Pro at $20 per month, Max 5x at $100 per month, and Max 20x at $200 per month.
About the Author
