Claude Code API Usage Limit Unexpectedly Reduced on MAX Plan

Same situation, I am coding with the claude code Max plan within its limits since weeks and today claude code stopped working after some medium sized refactoring and planning tasks with the message "Claude usage limit reached. Your limit will reset at 12am (....)" Previously I could easily handle those tasks within the limit.

Platform: Windows, WSL

+1

+1, is this a bug? or a tweak? soooo bad...

I am having the same problem. I have never hit the limit before on the max plan and now its saying I have to wait 2 hours for it to reset? I have never ever gotten close to maxing the usage. They silently reduced the limit. I want what I payed for!!! Not nerfing the product once we are on it!!!

+1
Same for me, I thought version 1.0.52 would solve this problem, but no. I got the MAX plan to avoid this situation, and after updating to version 1.0.51, it started to happen. Terrible, I made like two "plan mode" with OPUS, and I received the message already. I use OPUS only to plan; for the development, I use the Sonnet model.

Even using only the Sonnet model, the limit was reached too fast.

Platform: Windows, WSL

It's dumbed down as well. Im running circles around a problem....

same for me , and to be honest, sonnet is not really usable .... liar and cheater , so will try Kimi K2 today ... done with claude code

Claude Code was working great for a while, but I am having the same issues quite frequently. They started happening periodically a week ago, the context windows being chewed up out of no where (yes my context settings are already normal in settings.json), Claude forgetting instructions and destroying code (like blindly using sed when everywhere in the context/settings I reaffirm to NOT use that tool).

Today I was doing the simplest clerical fixes to explicitly avoid filling the context window, and low and behold, I have used up 5 hours of context in 1 hour editing files...

Seriously becoming difficult to maintain a workflow when the times Claude chooses to start clocking the windows are inconsistent as well, very disappointed. 😢

EDIT: It's been mentioned elsewhere but just in case this helps anyone, intentional use of specific models for specific tasks (i.e; Opus for planning, refactoring and code-mapping, and then using Sonnet for following a well structured plan, and specific general coding operations) using /models in the Claude Code CLI has improved the experience and token usage for the time being.

+1

+1

+1. Paying for MAX 100 dollars, reaching opus limit within like 3-4 prompts using already 20% of my 5 hours limit. Sonnet is good as well, but opus solves my complex problems. It just does not feel I can even use opus...

I'm experiencing the same issue! It feels like I'm on the Pro plan now. Before July, I hardly ever hit a daily limit. Now, after just two hours of coding, I've reached it! Even the Pro plan felt more generous before.

Alright fellas, looks like the gig is up. npx ccusage revealed we were all using $200+/DAY usage. They've caught on, and the YouTubers make it worse with all the videos talking about how good it was. This reminds me of moviepass, if price mismatch exists, users will definitely take advantage. Now we will need to invest in our own infrastructure and start using a fine-tuned open source model. I don't see Anthropic reversing this.

+1 The absence of data in the new Dashboard makes it impossible to determine usage levels.

+1 feel like using Pro Plan even in Max account

+1 Today, i started fresh session(i used claude or claude code lastly 14 hours ago). But i hit the opus 4 limit with just 1 prompt...

ps. I'm using max plan

+1

+1

+1

Check new rate limits from today's announcement: https://x.com/AnthropicAI/status/1949898502688903593
That should explain some of the "issues" encountered.

claude pro is barely usable, maybe 1 hour of coding before hitting the 5 hr limit with a small project. I get interrupted countless times in between some task. now with the new limits in place, lets see if i will also hit daily and weekly limits, lol, then i won't continue the subscription for sure. they present it like they are getting abused by some power users, but actually the very normal usage is already extremely limited.

Claude Code Opus on fresh session with extended thinking had to be manhandled into producing a basic design spec for one small module. Then proceeded to immediately implement it by ignoring the design. When asked about it: "You're absolutely right! Design documents are an important part of the Enterprise Software Design Lifecycle (SDLC). Would you like me to help you create a design document?". /Cancels subscription to Claude.

Just switched to Gemini CLI - their flash model outperforms Opus now, and you get almost unlimited usage on the T1 or free tier (sign up for an API key but DON'T put billing info). It respects rules, does what I want, has a long context, is very fast, doesn't spam emojis everywhere, and updates documents and cross-linked code files perfectly. 6+ hours and I have not hit a rate limit. Did I mention it's free? 99.9% of the problems I had with Claude are gone. Never going back unless someone at Anthropic can guarantee a service level for both usage and quality.

+1

+1
when https://status.anthropic.com/ experiences issues, Claude Code quickly encounters limitations

+1000

Having such limits makes it unusable. Similar to when GitHub Copilot had to adjust their premium request counting...

this usage limit makes Claude completely unusable for me. Glad i paid monthly instead of annually - I'm done with it this month. ChatGPT is worse but at least I don't get locked out after ~20-30 minutes of light/moderate use.

This issue has been automatically locked since it was closed and has not had any activity for 7 days. If you're experiencing a similar issue, please file a new issue and reference this one if it's relevant.