AI company Anthropic has announced major changes to how users can access Claude Fable 5, one of its most advanced AI models. Starting July 20, the model will no longer be available under the same conditions across all paid subscription plans. The company says the changes are designed to optimize the use of its computing resources while prioritizing premium customers.
Under the new policy, Max and Team Premium subscribers will continue to have access to Claude Fable 5. However, usage of the model will now be limited to 50% of their overall subscription quota, introducing stricter controls over resource consumption.
Meanwhile, Pro and Team Standard subscribers will lose subscription-based access to Claude Fable 5. Instead, Anthropic will provide these users with a one-time $100 credit, which can be used to access the model through a pay-as-you-go pricing system based on actual usage.
According to Anthropic, the decision reflects the growing cost of operating frontier AI models. Claude Fable 5 requires significant GPU capacity and computational infrastructure, making it one of the company’s most resource-intensive models. As demand for advanced AI continues to grow, balancing performance with infrastructure costs has become increasingly important.
The company had initially considered removing Claude Fable 5 from standard subscription plans entirely. However, after reviewing user feedback and market demand, Anthropic revised its approach and decided to keep the model available for its highest-tier subscribers.
The update also highlights a broader trend across the AI industry. Competition is no longer focused solely on building the most capable models—it is increasingly about managing the infrastructure required to run them efficiently. GPU availability, computing power, and operational costs are becoming key competitive factors for leading AI companies.
As a result, more AI providers are adopting premium subscription tiers, usage limits, and consumption-based pricing models. Anthropic’s latest policy change is another sign that sustainable infrastructure management is becoming just as important as advances in model performance.
