Hidden Perk of Claude Subscriptions: Max Includes $100/month in API Credits

Claude Max/Ultra include monthly API credits, but you must manually link a Console organization under strict rules.
Claude's Max and Ultra subscriptions include $100 and $200 in monthly API credits respectively, but this money doesn't arrive automatically—users must manually link a Console organization via Link Organization in the API Credits section of the Billing page on Claude.ai. There are two hard requirements: you must be the subscriber yourself, and the subscription must be at least seven days old; the link is also irreversible, so choose your organization carefully. Credits can be used for the Claude API, Console playground, managed agents, and the agent SDK, but do not cover terminal Claude Code or in-app extra usage. Unused credits are reset at the end of each billing cycle, and when exhausted, requests simply stop without charging your subscription plan. This benefit is therefore better suited to users with ongoing API development needs.
Claude's subscription plans include a benefit many users overlook: Max comes with $100 in API credits per month, and Ultra with $200. But this money doesn't arrive automatically—you need to take the initiative to complete a one-time linking step.
Credits Don't Arrive Automatically
According to the video source, these API credits are a "subscription add-on" benefit, but only if you link your subscription account to a Console organization. The process isn't complicated: open Claude.ai in a browser, go to Settings, then Billing, find the API Credits section, and select Link Organization.

There are two hard requirements for linking: you must be the subscriber yourself, and the subscription must be at least seven days old. In other words, you can't claim it on the day you sign up—you have to wait a week. Once you meet the conditions, select an existing Console organization or create a new one, read the terms, and confirm the link.
What Is a Console Organization?
Anthropic Console is a management platform for developers, separate from the consumer-facing Claude.ai app. A Console organization is the billing and management unit for developers using the API: every API key, usage record, bill, and balance belongs to a specific organization. One account can create or join multiple organizations—for example, separated by project or by team. Subscription API credits can only be deposited into one organization, so the organization you select during linking is where the credits actually land. Understanding this is key—the credits aren't sent to your personal account but to the balance of the organization you specify, and subsequent API calls must use a key under that organization to draw on these funds.
You Only Get One Chance to Link
Here's an easy pitfall: the linking action is irreversible. As the video puts it, "you get one link and you can't change it yourself." So think carefully when choosing an organization.

Once linked, the credits appear in that organization's balance, with no credit card required.

What You Can Spend It On
The credits cover Claude API calls, the Console playground, managed agents, and the agent SDK. In other words, most development-related usage through the API channel can be offset with these credits.
But there are clear exclusions: Claude Code in the terminal can't use these credits, nor can in-app extra usage.

Another rule to keep in mind is the expiration mechanism—unused credits for the current period are forfeited at the next billing cycle and do not accumulate. So it's better suited to developers with ongoing API needs rather than for "grabbing it now to save for later."
Where Managed Agents and the Agent SDK Fit
The Agent SDK is a development toolkit from Anthropic that lets developers build agent programs in code that can autonomously call tools and perform multi-step reasoning; Managed Agents is a service model that hosts the running, orchestration, and session management of such agents on Anthropic's side, so developers don't need to build their own scheduling and state management infrastructure. Both go through the API channel and are billed by token or runtime, so they can be offset by the included credits. In contrast, Claude Code in the terminal is a separate subscription-based product whose usage is covered by the subscription plan itself and doesn't go through the above billing, which is why it can't use these credits.
What Happens When Credits Run Out
The design here is fairly restrained: once credits are exhausted, requests simply stop, but it won't start charging your subscription plan instead. In other words, it's a hard cap—there's no risk of "overspend billing."
For developers who frequently use the API for prototypes and agent experiments, these included credits effectively raise the value of the subscription. The one core question in deciding whether it's worth linking: will you actually build things with the API? If you only use it for everyday chat, the practical value of this benefit is limited.
One-line Summary
Claude Max/Ultra include monthly API credits, but you must manually link a Console organization under strict rules.
Paragraph Summary
Claude's Max and Ultra subscriptions include $100 and $200 in monthly API credits respectively, but this money doesn't arrive automatically—users must manually link a Console organization via Link Organization in the API Credits section of the Billing page on Claude.ai. There are two hard requirements: you must be the subscriber yourself, and the subscription must be at least seven days old; the link is also irreversible, so choose your organization carefully. Credits can be used for the Claude API, Console playground, managed agents, and the agent SDK, but do not cover terminal Claude Code or in-app extra usage. Unused credits are reset at the end of each billing cycle, and when exhausted, requests simply stop without charging your subscription plan. This benefit is therefore better suited to users with ongoing API development needs.
Related articles

AI Tools Quietly Altering Settings: Postman Silently Installs Plugin, Codex Sharing Option Auto-Resets
AI Daily reports: Postman silently installs a Mac plugin and self-approves startup hooks, logged as auto-existing upgrade; Hacker News developers say Codex data-sharing opt-out flips back. Plus Simon Willison's raw float weight experiment with QEN 3.8-27B.

AI Learns to Rewrite AI: MIT's "Weightspace Programming" Redraws the Path to Recursive Self-Improvement
MIT team proposes weightspace programming: GPT-6 Astra reads and edits AI model weights directly. Math scores jump several-fold, hand-written networks beat trained models. Eight experiments analyzed, plus Anthropic's same-day warning and the real boundaries of recursive self-improvement.

Running Qwen 3.8 on a Laptop to Read Raw Weights: 46 Minutes Per Answer
Simon Willison reproduced weight space programming with Qwen 3.8 27B on a laptop: a 17GB model file, 46 minutes per answer, 43,506 reasoning tokens, Pong over an hour, Freeway failed. Suited for interpretability researchers.