Good news, OpenClaw fans – you can once again use your Claude AI subscription to power the hit, open source, autonomous AI agent trailer! But there’s a big catch to how it’s perceived.
A few hours ago, Anthropic announced via the official developer communication account on X, @ClaudeDevsClaude changed paid subscription levels, introduces a new subcategory "Agent SDK" loans for all paid subscribers, they can now book separately "programmed" It uses including external third-party agents such as OpenClaw.
The move is a big departure from Anthropic The policy implemented at the beginning of April 2026 Anthropic has publicly banned the use of AI subscriptions to power such non-Anthropic agents and trailers after it said it caused capacity and service issues.
The problem was that some Claude subscribers were paying $20-$200 per month under Anthropic’s Claude Pro and Max subscriptions, but were consuming hundreds or even thousands of dollars worth of tokens (units of data) at those prices through OpenClaw (and similar autonomous) agents. This was an unsustainable position for Anthropic’s finances and limited computing infrastructure to get the models out to end users.
To be clear, even though it enforced its old ban against OpenClaw and similar agents last month, Anthropic never completely cut off Claude’s ability to use OpenClaw. Instead, it directed users to pay through the company application programming interface (API)billed per usage (priced per million tokens rather than a fixed monthly rate as subscriptions offer) or pay additional usage credits on top of their subscriptions.
Anthropic now provides another way for Claude subscribers to use their subscription account to pay for third-party agents.
However, the recovery comes with a significant cap: programmatic usage is no longer subsidized by the overall subscription pool, but instead is limited to a fixed, fixed monthly credit of $20-$200, depending on your Clod plan, and billed at API rates.
In other words, if you don’t use these new Agent SDK credits, they simply expire at the end of the month. If you use them all, you won’t be able to access the overall subscription usage limits to cover any additional usage – you’ll need to purchase additional usage credits instead.
Why did Anthropic block Claude subscriptions from OpenClaw (and other third-party agent AI plugins) in the first place?
To understand why this recovery is important, you need to look at the technical friction that led to the original ban on April 4, 2026.
Built to maximize first-party tools like Anthropic’s Claude Code and Claude Cowork. "operational cache hit rates"— a method of reusing previously processed text to save expensive computing cycles.
Third-party tools like OpenClaw, which allow users to run autonomous agents through external services like Discord or Telegram, are often not optimized for these efficiencies. The head of Claude Code, Boris Cherny, noted that these third-party services "It is very difficult for us to do it sustainably" because they bypassed the caching mechanisms that allowed Anthropic to offer flat-rate subscriptions.
The large volume of data being processed by inefficient agents threatened the stability of the system for a wider user base. Even with Anthropic’s massive expansion into new hardware, incl 300MW Colossus 1 data center and its 220,000+ GPUs – demand for agent workflows was consistently outstripping supply.
New "Agent SDK credit" the system solves this technical bottleneck by passing the cost of the inefficiency back to the user. By introducing a specific dollar amount of credit, Anthropic no longer has to "eat the difference" in unoptimized third-party code. If an agent is inefficient and burning tokens, it drains the user’s new $20-$200 Agent SDK credit budget more quickly, rather than exceeding the cost of Anthropic’s fixed monthly subscription tiers.
Anthropic’s new programmatic loan system
Third-party access recovery is split across Anthropic’s compute tiers, creating a new hierarchy. "program power." How much Anthropic gives each user in terms of new Agent SDK credits (Claude Code, Claude Cowork, etc.)
|
Plan |
Monthly, Custom Agent SDK Credit (in addition to existing subscription plans) |
Context of Use |
|
Pro |
20 dollars |
Custom scripts and lightweight SDK usage. |
|
Maximum 5x |
100 dollars |
Medium agent automation. |
|
20x max |
200 dollars |
Professional-grade development environments. |
|
Team (Premium) |
$100/seat |
Automation of collective command. |
|
Enterprise (Premium) |
$200/seat |
Seat-based high-scale enterprise use. |
A sharp distinction is drawn between this system "interactive" and "programmed" workflows. If you chat with Claude in the browser or use Claude Code in the terminal to write interactive code, you still use your standard, high-capacity subscription limits.
as Anthropic technician Lydia Hallie wrote a post on X, "Just to add some clarity: you don’t pay extra. Same subscription, same monthly price." Hallie also included the following helpful diagram of how the new Agent SDK credits work:
But the moment you use it claude -p command for non-interactive tasks, run a GitHub Action, or integrate a third-party tool like OpenClaw, the system switches to a dedicated Agent SDK credit.
Once the Agent SDK credit limit ($20 for Pro plans, $100 for Max 5X, etc.) is used up, app usage stops unless user activates. "additional use" billed at standard, pay-as-you-go API rates.
Mainly, it’s a hard cap for those who consider the original subscription model to be an infinite resource. Credits don’t roll, ie "use it or lose it" the nature of the system forces the developer to reset their budget monthly.
Strategic implications
The licensing implications of this move are enormous "agent" ecosystem.
By explicitly allowing third-party applications such as Conductor and OpenClaw to authenticate through the Agent SDK, Anthropic legitimizes the workflow it previously tried to block.
But by doing this, he ended his period "settlement arbitrage".A $20 Pro subscription through OpenClaw can be used to manage agents that would cost hundreds of dollars with a standard API key in early 2026.
By moving to scalable lending, Anthropic aligns its subscription model with its Developer Platform (API). While offering this a "for free" buffer for subscribers, it enables the transition from high-volume, production-level automation to predictable, token-based billing.
This protects the company’s margins while still offering "sandbox" for developers to test without the immediate burden of an API-first account.
Community reactions are perhaps surprisingly negative
While anthropic executives frame the update as one "simplification"the developer community mostly noted this as a significant reduction in the cost of their subscriptions. The backlash focuses on the stark difference between the previous effective use and the new, measured reality.
Popular AI YouTuber and developer Theo Browne (@theo) T3.gg warned developers that this change represents a major devaluation for those using external tools. "If you use any of the following with Claude sub, your usage should be reduced by 25x," Theo mentioned T3 Code, Conductor, Zed and Jean as affected platforms. He concluded with a sharp warning: "They disguise it as “free credit”. Don’t fall for it".
Kun ChenMeta, a solo founder and former L8 engineer at Microsoft and Atlassian, interpreted the move as a complete surrender of Anthropic’s market leader. "is official. Anthropic has disabled ALL programmatic use of the claude subscription," Chen wrote and added that he found himself "Growing up about OpenAI" as a result. Chen defended it "Anthropic’s only lead was in coding, and gpt 5.5 has already changed that," signaling a potential migration of elite developer talent.
Other developers questioned the practical utility of the proposed loans. Ben HylakThe co-founder and chief technology officer of artificial intelligence agent observability and management startup Raindrop.ai expressed concern about the sustainability of Anthropic’s infrastructure. "this is either really stupid or shows how bad an anthropic point is :gpus," Hylak noted that before asking users openly for "Guess how much $20 turns in API credits".
The frustration extended to the marketing of the change. NeverThe founder of inkstone.uk said he did not believe the policy was drafted. "Wait what?! Have you given up more ways to use the subscription I paid for?! And you dare to pretend it’s a victory?". The sentiment highlights a growing rift between Anthropic and its power user base, who feel that previously included features have been stripped away under one name. "improve."
Bottom line for Anthropic subscribers and AI builders
Anthropic "restoration" is a tactical move to retain developers while strictly managing the physical limitations of computing. Until June 15 "agent" It will be period-sized for Claude subscribers.
The company successfully regained control of its margins, albeit at the cost of some of the goodwill of its most vocal users.
For an individual developer or enterprise AI builder relying on Anthropic models for OpenClaw, this is clearly an improvement over last month’s ban.
