Anthropic imposes weekly limits on Claude use, developers outraged

From August 28, the service will limit access to Claude to its most active users, citing system overload due to abuse.

Anthropic has announced weekly limits on the use of its AI platform Claude. The new limits will begin on August 28 and will add to the existing 5-hour limits. As the company notes, the restrictions will affect about 5% of users, but have caused discontent even among those who aren’t heavy users.

Anthropic’s new limits will affect about 5% of users, but have caused discontent even among those who aren’t heavy users.

The reason is excessive load from Claude Code users

According to Anthropic, most of the abuse is related to the Claude Code product, an AI tool for programmers. Some users were running Claude in the background 24/7, while others were sharing account access or reselling access to third parties. All of this, the company said, was negatively impacting system performance for everyone else.

“Claude Code has faced unprecedented demand since launch. We’re seeing policy violations ranging from account sharing to overselling access. This degrades the service for the rest of us,” Anthropic said in a statement to VentureBeat.

The company also clarified in an email to subscribers that the behavior of individual users leads to a decrease in the overall stability of the platform. To keep the service running reliably, Anthropic has decided to temporarily enforce weekly limits while promising to find a more sustainable solution for “long-term usage scenarios.”

What exactly is being limited and how much can be used

Although specific limits have not been officially named, Anthropic says that most users of the Claude Max 20x plan will be able to count on:

• 240 to 480 hours of Sonnet 4 usage,
• 24 to 40 hours of Opus 4 use per week.

Specialized users who run multiple instances of Claude Code at once may reach the limit faster.

Those who exceed the weekly limits will be offered to continue for an additional fee at standard API rates. Also, some Anthropic enterprise customers may have special terms without limits, but regular subscription users will now have to take the weekly limits into account.

Anthropic’s corporate customers will also have special terms without limits, but regular subscription users will now have to take the weekly limits into account.

Community reaction: from discontent to understanding

The restrictions have sparked a wave of criticism among developers. Many feel it is unfair that conscientious users suffer because of the actions of a minority. Those who use Claude for long-term projects or experimenting with autonomous agents are particularly unhappy.

Nonetheless, some of the community has been sympathetic to the changes, noting that Anthropic has likely faced the challenge of keeping the service stable in the face of surging workloads.

Context: competition and scaling

Anthropic is actively developing the Claude platform: in June, the service became a no-code tool and also got an enterprise version for the financial sector. However, growing popularity requires additional computing resources. Unlike Google, which is phasing out limits on some models, Anthropic and OpenAI are sticking with a pay-per-access tiered model, where large amounts of capacity are made available for a fee.

The story Anthropic introduces weekly limits on Claude usage, developers outraged was first published on itzine.ru.