If you’re a hopeless Claude user like me, be forewarned. New weekly rates are here. See email at end of this post.
There’s nothing worse than getting rate-locked right before a deadline. You’re in the zone, Claude’s cranking out code and then — BAM — you hit the wall. “Please wait a few hours,” it says, before your rate resets. Meantime, your project(s) come to a standstill.
It’s frustrating, sure. But it also speaks to just how indispensable Claude has become. When you’re cursing at rate limits, you’re really acknowledging that this AI has become your digital co-pilot. And now, Anthropic’s tightening rates even further by adding a weekly limit.
The Great Rate Reckoning of 2025
Starting August 28, Claude subscribers are getting new weekly limits on top of the existing 5-hour caps. Why? Turns out some users have been treating Claude like a 24/7 code sweatshop, running it in the background around the clock. Others have been sharing accounts or reselling access. Digital scalping, if you will. Sigh, and now it’s class detention time thanks to a few rotten apples.
Here’s what’s changing:
For most users (95%, according to Anthropic), this won’t matter. But for the power users? Expect roughly:
- 240-480 hours of Sonnet 4 per week
- 24-40 hours of Opus 4 per week
Heavy Opus users with massive codebases or those, like me, running multiple Claude Code instances? You’ll hit the ceiling faster.
Juggling AI Tabs
Here’s my confession: whenever I get Claude-walled, I perform what I call the AI tab shuffle. Claude locks up? Over to ChatGPT. ChatGPT starts hallucinating? Hello, Gemini. It’s like dating apps, but for LLMS. Worst case scenario: I have ollama on my UNRAID with a handful of surprisingly good (and free) SLMs.
The real pain comes when Claude — my go-to for lengthy scripting tasks for the Stark Insider server — goes on its mandatory break. That’s when all code development grinds to a halt. Sure, the others can help, but I find Claude to be the best at bash scripting.
The Wild West
It’s boom time!
Anthropic’s email reveals the dirty secret of the AI boom: people are absolutely hammering these systems. Running Claude 24/7 in the background? Sort of reminds me of running servers night and day to mine for crypto.
I also wonder if Anthropic is just flat out struggling to scale, all while maintaining some sort of realistic path to profitability, or at least trying to lessen losses? I tend to notice Claude struggling with more connection errors than, say, ChatGPT or Gemini. And those latter two are much larger operations with Google, of course, having decades of experience with massive infrastructure.
Try installing ollama in a Docker container, and install a few models (phi3:mini, deepseek-coder:1.3b and llama3.2:3b are impressive!). You’ll quickly learn first hand how much GPU it takes to run these things. My server can barely run med.-sized models, let alone large ones. Many of the small ones are fine. And it’s pretty cool that you can even run these privately (no APIs required) in your own home. But it does remind me how much hardware and processing power companies need to continually invest to keep operations running.
What This Means for the Rest of Us
For the 95% who won’t notice these changes, carry on. For the 5% who will, welcome to the new world of AI resource management. Start thinking of your Claude usage like a budget. For me, that means being more selective when I use Claude. If it’s something not related to critical reasoning and coding. I might just get ChatGPT to work on it. Another thing I’ve (foolishly) learned: don’t pasting in 1,800 lines of code into Claude and not expect to hit your rate sooner than later. Hint: modularity!
P.S. – Anthropic also acknowledged the recent reliability issues. Nothing like announcing rate limits while your service is having a rough patch. Bold move, but at least they’re being transparent about it.
Important updates to your Max account usage limits
Hi there,
Next month, we’re introducing new weekly rate limits for Claude subscribers, affecting less than 5% of users based on current usage patterns.
Claude Code, especially as part of our subscription bundle, has seen unprecedented growth. At the same time, we’ve identified policy violations like account sharing and reselling access—and advanced usage patterns like running Claude 24/7 in the background—that are impacting system capacity for all. Our new rate limits address these issues and provide a more equitable experience for all users.
What’s changing:
Starting August 28, we’re introducing weekly usage limits alongside our existing 5-hour limits:
- Current: Usage limit that resets every 5 hours (no change)
- New: Overall weekly limit that resets every 7 days
- New: Claude Opus 4 weekly limit that resets every 7 days
- As we learn more about how developers use Claude Code, we may adjust usage limits to better serve our community.
What this means for you:
- Most users won’t notice any difference. The weekly limits are designed to support typical daily use across your projects.
- Most Max 20x users can expect 240-480 hours of Sonnet 4 and 24-40 hours of Opus 4 within their weekly rate limits. Heavy Opus users with large codebases or those running multiple Claude Code instances in parallel will hit their limits sooner.
- If you do reach a weekly usage limit, you’ll have the option to purchase more usage at standard API rates to continue working without interruption. This is completely optional.
- You can manage or cancel your subscription anytime in Settings.
We take these decisions seriously. We’re committed to supporting long-running use cases through other options in the future, but until then, weekly limits will help us maintain reliable service for everyone. Max 20x subscribers can purchase additional usage at standard API rates if needed.
We also recognize that during this same period, users have encountered several reliability and performance issues. We’ve been working to fix these as quickly as possible and will continue addressing any remaining issues over the coming days and weeks.
–The Anthropic Team