Skip to main content
Back

Claude Code usage limits are changing: 17% drop incoming, but still net positive

Anthropic raises Claude Code limits 25%, Grok shops online, Claude fixes misalignment in larger models, fired devs build AI CEO

August 31, 2026 6 min read

Hey,

AI agents are quietly crossing a line: from thinking to spending. Grok Bot just got a credit card. Claude Code is tightening its belt. Both signal the same thing: the agentic era has real-world costs, and labs are figuring out who pays.

Anthropic's limit shuffle is peak corporate math. A "25% raise" that feels like a 17% cut. We see you.

Meanwhile, Pangram's 99.99% true negative rate on AI detection quietly matters more than it sounds. As agents write more, knowing what's human becomes infrastructure.
In Partnership with · partner_image
Author photo
Today's Author
Lior Alexander. Founder of AlphaSignal and former ML Engineer. Previously built ML systems at Iguazio, Guesty, Enphase, Mila.
Summary

Read time: 6 min 25 sec

Top News

▸ Anthropic permanently raises Claude Code weekly limits by 25% starting September 14

OpenRouter

▸ One API For Every Model

Top News

▸ Grok Bot can now shop online and pay using single-use cards

Teleport

▸ What did your AI agent do while you were away?

Top News

▸ Claude Code desktop app now lets you resume CLI sessions mid-conversation

Signals

1. Anthropic lets Claude autonomously fix misalignment in models up to 4.7x larger
2. Granola: Walk out of in-person meetings with notes done. Granola on Apple Watch: tap your wrist and focus on the person you're meeting
3. Pangram Labs ships AI text detector with 99.99% true negative rate
4. New video pretraining method matches Meta's V-JEPA 2 at up to 20x less compute
5. Toby Ord's new paper shows why a true AI intelligence explosion is mathematically harder than assumed
6. Fired developers build open-source AI CEO using 8 Claude agents
Top News
Anthropic permanently raises Claude Code weekly limits by 25% starting September 14
21,788 Likes
Anthropic permanently raises Claude Code weekly li

Heads up: Anthropic just changed the usage limits for Claude Code, and the math is a little confusing at first glance.

Here is what is actually happening. There was a temporary 50% boost to weekly limits running right now. That boost ends soon. After that, limits settle at a permanent 25% increase over the original baseline.

So compared to where limits are today, you will feel a 17% reduction. But compared to where limits were before the temporary boost, you are still getting more. Net positive, just not as generous as the temporary spike.

This affects all paid plans:

  • Pro, Max, Team, and seat-based Enterprise

The good news is Anthropic says they are working on changes that give you better visibility into your usage and more control over it, so you are not flying blind when limits get tight.

For now, plan your Claude Code sessions around the new permanent ceiling.

Read more
forward →
Presented by OpenRouter
One API For Every Model

OpenRouter gives you access to 500+ AI models from 80+ providers through one API.

Switch models without rebuilding your stack, route requests based on price, performance, and availability, and stay resilient with automatic failover when providers go down.

Get the flexibility to use the right model for every request - without juggling provider keys or managing complex routing infrastructure.

Start building now
partner with us →
Top News
Grok Bot can now shop online and pay using single-use cards
10,983 Likes
Grok Bot can now shop online and pay using single-

Your AI just got a credit card. Grok Bot can now shop online for you, and it is actually pretty clever about how it works.

Here is the setup: connect the Stripe Link plugin inside Grok Bot's Plugins section, then tell your Bot what to buy. It searches products, compares options, builds the cart, and heads to checkout on its own.

Now the part everyone asks about: is it safe?

  • Grok Bot never sees your real card number
  • Stripe Link creates a single-use virtual card for each purchase, then throws it away
  • You approve every spend request before any money moves, seeing the merchant, amount, and full breakdown
  • You can set rules: allowed sites, budget limits, what it must never buy

Someone already used it to order a Tesla Model Y. So yes, the ceiling is high.

Currently US only. Mobile approval coming soon.

Read more
forward →
Presented by Teleport
What did your AI agent do while you were away?

Once agents run in production, you need to know what they did, whether it was expected, and what risk comes next.

Teleport helps you:

  • Turn 3-hour sessions into 30-second summaries
  • Catch workarounds with Agentic Classifiers
  • Triage activity with Risk Scoring

Join Teleport's Ben Arent on September 17.

Register now
partner with us →
Top News
Claude Code desktop app now lets you resume CLI sessions mid-conversation
7,297 Likes
Claude Code desktop app now lets you resume CLI se

Ever start a coding session in the terminal, then wish you could move it to a proper app with a nicer interface? That gap just closed.

Claude Code now lets you resume any terminal session inside the desktop app. Just type /resume in the app, pick the session you want, and it loads up with the full conversation history and context, exactly where you left off.

Why does this matter? Context is everything when coding with AI. Losing it means re-explaining your whole project from scratch. Now you never have to.

  • Start a session in the terminal, finish it in the app
  • Full conversation history carries over, nothing gets lost
  • One command: /resume

This opens up a smoother workflow where you can jump between environments freely. Terminal for speed, desktop app when you want more comfort. Same brain, different window.

Read more
forward →
Signals
1
Anthropic lets Claude autonomously fix misalignment in models up to 4.7x larger
4,773 Likes
2
Walk out of in-person meetings with notes done. Granola on Apple Watch: tap your wrist and focus on the person you're meeting
Presented by Granola
3
Pangram Labs ships AI text detector with 99.99% true negative rate
695 Likes
4
New video pretraining method matches Meta's V-JEPA 2 at up to 20x less compute
675 Likes
5
Toby Ord's new paper shows why a true AI intelligence explosion is mathematically harder than assumed
317 Likes
6
Fired developers build open-source AI CEO using 8 Claude agents
1,292 Likes
At Alpha Signal, our mission is to build a sharp, engaged community focused on AI, machine learning, and cutting-edge language models, helping over 300,000 developers stay informed and ahead. We're passionate about curating the best in AI, from top research and trending technical blogs to expert insights and tailored job opportunities. We keep you connected to the breakthroughs and discussions that matter, so you can stay in the loop without endless searching. We also work closely with partners who value the future of AI, including employers and advertisers who want to reach an audience as passionate about AI as we are.

Our partnerships are based on shared values of ethics, responsibility, and a commitment to building a better world through technology.Privacy is a priority at Alpha Signal. Our Privacy Policy clearly explains how we collect, store, and use your personal and non-personal information. By using our website, you accept these terms, which you can review on our website. This policy applies across all Alpha Signal pages, outlining your rights and how to contact us if you want to adjust the use of your information. We're based in the United States. By using our site, you agree to be governed by U.S. laws.
Looking to promote your company, product, service, or event to 300,000+ AI developers?
Work with us