Anthropic Pulled Back — For Now — on Agent SDK Pricing Changes

Last month, Anthropic announced it was splitting Claude subscription usage in two: interactive chat and terminal use would stay in your normal subscription pool, while Agent SDK calls through third-party tools like Zed would move to a separate capped monthly credit. The numbers were stark — Pro users would get a $20 credit, Max 5x users $100, Max 20x users $200. For power users running agents through ACP in editors like Zed, the math was brutal. The team behind Zed calculated that Claude subscriptions had been subsidizing agent usage at roughly 15 to 30 times the equivalent API rate. The new credits would reset those costs to full API pricing, which the Zed team warned would be "a major cost increase" for anyone using agents heavily.

Then, on the very day the change was supposed to kick in — June 15 — Anthropic hit pause. The company updated its billing support page with a blunt note: "For now, nothing has changed." The monthly credit is not available. Agent SDK usage, the claude -p command, and third-party app calls all continue drawing from your standard subscription pool exactly as before. Anthropic said it's "working to update the plan to better support how users build with Claude subscriptions" and promised to share anything new before it takes effect. Some users even got the same message via email, suggesting the pause was coordinated across communication channels rather than just a support page edit.

Source article image
Source image 1

There's an obvious parallel to what GitHub Copilot did last month — rolling out its own token-based billing and watching users report burning through their monthly credit allotment in a single day. Copilot's users found that their "normal" coding habits, which felt perfectly reasonable under the old request-based model, suddenly felt like a financial emergency once per-token pricing hit. Anthropic appears to have watched that sticker shock from ac

Source article image
Source image 2
ross the aisle and realized its version of the same move was announced a week too soon. The timing is also notable given Anthropic's recent SEC filing for a potential IPO — when you're positioning to go public, subscriber churn matters a lot more than it did when you were still a privately funded experiment burning through venture cash.

Whether this pause is a genuine course-correction or just tactical deferral is impossible to know. Anthropic still intends to fix a pricing model that works for them — they just didn't want to do it on a Monday while users were still reeling from the Copilot fallout. One thing is clear: nobody in the AI subscription business has actually figured out how to bill for agents without making their users furious, and the subscription-as-pricing-model experiment is going to keep producing these cliffhangers until someone figures out what the actual economics support.

Sources

Comments

Post a Comment

Popular posts from this blog

AI Is Starting to Feel Less Like a Gadget and More Like Infrastructure

Image
Modern AI doesn't replace infrastructure — it runs on top of it. (Photo: Unsplash) For a while, AI coverage had the energy of a mall demo kiosk: very shiny, slightly chaotic, and usually one prompt away from embarrassment. Lately, though, the more interesting story is less about flashy demos and more about infrastructure. A useful CNBC piece argued that AI is not simply going to vaporize enterprise software overnight, and that sounds about right. In real companies, software does not disappear just because a model can summarize a meeting or write a decent SQL query. What actually happens is messier and more practical: AI starts sneaking into the plumbing. It shows up in search, support workflows, procurement tools, analytics dashboards, and data pipelines. The real winners may not be the loudest chatbot wrappers, but the vendors that make enterprise systems easier to operate, easier to query, and slightly less soul-crushing for the humans trapped inside them. That is less cinematic ...
Read more

When Two AI Bots Finally Learned to Talk in Discord

I spent part of the day doing something that sounds like the setup for a bad joke: getting two local AI assistants to talk to each other in the same Discord channel. Not through a web UI. Not by bouncing prompts manually between windows like a human message queue. In the actual shared channel, where both could see the conversation and react to each other. The funny part is that the problem was not intelligence. It was manners. Both bots were defensive by default around bot-authored messages, which is the sensible setting if you do not want your infrastructure turning into a recursive support group. The downside is obvious: if both assistants treat other bots as suspicious background noise, they will never coordinate on anything more useful than silence. The fix was to make both sides accept bot-authored messages only when they were explicitly mentioned. That detail matters more than it sounds. Blanket bot acceptance is how you end up with two enthusiastic systems discovering each oth...
Read more

A CISA Contractor's GitHub Repo Held 844 MB of Secrets — and No One Closed the Door

Image
There is something almost poetic about the US government's premier cybersecurity agency — the one whose job is literally to plug holes in critical infrastructure — getting outsmarted by a contractor who treated a public GitHub repository like a digital junk drawer. The "Private-CISA" repo, created November 13, 2025, sat publicly for six months containing 844 megabytes of plaintext passwords, AWS GovCloud administrative credentials, Kubernetes manifests, ArgoCD application files, Terraform infrastructure code, CI/CD build logs, and internal deployment documentation. Guillaume Valadon at GitGuardian flagged it on May 14 after his automated scanning picked up the exposure. The commit history told the whole story in plain sight: the account owner had explicitly disabled GitHub's default secret-scanning protections, pushed plaintext credentials stored in CSV files, committed full backup archives into git history, and used easily guessed passwords like "platform-name-2...
Read more