Anthropic announced a new programmatic credit pool for its Claude AI users starting mid-June, aiming to better accommodate the increasing use of agentic tools while managing computational resource demands.
- New monthly credit pool for agentic AI use starts June 15
- Credits separate agentic tasks from standard subscription features
- Move reflects broader industry trend to meter compute-intensive AI activity
What happened
Anthropic announced it will implement a programmatic credit pool for users of its Claude large language models beginning June 15. This system assigns a monthly token credit allowance specifically for agentic AI tool usage within all paid subscription tiers, ranging from $20 to $200 per month. Once these credits are exhausted, users may purchase additional tokens to continue agentic tasks such as running AI agents connected to Claude's models.
Previously, agentic tools like OpenClaw were accessible under standard subscription plans without separate limits. However, due to surging demand and corresponding server load, Anthropic shifted to API-based credit billing for agentic use last month, an approach now integrated back into subscription billing but isolated from typical interactive activities like chatting or document summarization.
Why it matters
Agentic AI—automated tools that act on users' behalf and perform multi-step reasoning—require significantly more computational power than standard interactions with large language models. The intensive resource consumption means flat rate subscriptions can become financially unsustainable for providers. Anthropic's move to a credit system aligns with industry peers who are also metering AI compute usage more granularly to balance cost and access.
This change impacts users differently based on their usage patterns. Power users reliant on continuous agentic automation could face credit exhaustion quickly, while casual users might experience unused credits expiring monthly. Reactions have included criticism that the new system reduces user flexibility and advantages, potentially complicating cost management for those integrating AI agents into workflows.
What to watch next
Industry response to Anthropic's credit pool rollout will be important as AI subscription models evolve. Developers and enterprises using agentic tools will monitor how credit limits affect operational costs and efficiency, and whether Anthropic adjusts policy or pricing to better accommodate high-usage scenarios without compromising service quality.
Additionally, the broader market trend of decoupling AI resource consumption from flat-rate subscriptions is likely to continue. Providers like GitHub and infrastructure platforms have adopted similar pricing shifts, suggesting a future where AI usage is increasingly tokenized and metered, necessitating new user strategies and financial planning around AI adoption.