Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Pro Max 5x quota exhausted in 1.5 hours despite moderate usage

github.com

Post reply on HN
Print view View original

351–360 of 695 posts

Re: Pro Max 5x quota exhausted in 1.5 hours despite moderate usage

#351

Post by stavros »

brunooliv wrote: post #321
stavros wrote: post #146

Earlier quoted context omitted.

> Anthropic CAN change their limits and rates as they see fit, there’s never been hard promises or SLOs on these plans. No they can't. When I buy an annual subscription and prepay for the year, they can't just go "ok now you get one token a month" a day in. I bought the plan as I bought it. They can't change anything until the next renewal.

It's the nature of SaaS software, right? It doesn't need to be an enforced "hard change", but, let's say that they trained Opus 4.6 to be more "verbose" or to explore more files to gain more context for it's own tasks. If your limits stay "the same", but you then use Opus 4.6, your quota will be exhausted much faster, it's just how it works. Note that some features are simply NOT made for these Pro, Max, Max 5x or wh…

What you say makes sense, but they very actually reduced the token limits. We had, say, 20M tokens/week before, now we have 18M tokens/week (example numbers). They didn't just make a model that eats tokens faster.

Re: Pro Max 5x quota exhausted in 1.5 hours despite moderate usage

#352

Post by stavros »

maerF0x0 wrote: post #277
stavros wrote: post #146

Earlier quoted context omitted.

> Anthropic CAN change their limits and rates as they see fit, there’s never been hard promises or SLOs on these plans. No they can't. When I buy an annual subscription and prepay for the year, they can't just go "ok now you get one token a month" a day in. I bought the plan as I bought it. They can't change anything until the next renewal.

That probably is somewhere in the EULA or other contract you agreed to. I'm not arguing it's any kind of fair, nor am I a lawyer so IDK if it's enforceable, but I bet it's in there somewhere.

Well yes, that's how they get you, but at some point the law will have to change to become fair.

Re: Pro Max 5x quota exhausted in 1.5 hours despite moderate usage

#353

Post by jatora »

gruez wrote: post #158
sunaookami wrote: post #124

Earlier quoted context omitted.

Set MAX_THINKING_TOKENS to 0, Claude's thinking hardly does anything and just wastes tokens. It actually often performs worse than without thinking.

Not the guy you're responding to, but when this happens the token counter is frozen at some low value (eg. 1k-10k) value as well, so it's not thinking in circles but rather not thinking (or doing anything, for that matter) at all.

i was having this issue yesterday. the same prompt would send it into a loop where it would appear to be doing nothing for 30+ minutes until i cancelled it. it would show 400 tokens used and thats it. I tested on a previous version (2.1.68) and it still ran into this neverending loop BUT at least the token count kept steadily increasing.

So we are seeing 1. some sort of model degredation is my guess (why it can't break a thinking loop on some problems), as well as 2. a clear drop in thinking token UI transparency

when i left it running overnight it finally sent a message saying it exceeded the 64000 output token limit

Re: Pro Max 5x quota exhausted in 1.5 hours despite moderate usage

#354

Post by 0xbadcafebee »

Please remember you do not need Anthropic. There are cheaper subscriptions with higher rate limits. Comparison of subscriptions to API: https://codeberg.org/mutablecc/calculate-ai-cost/src/branch/... Score/price comparison: https://benchlm.ai/llm-pricing

Opus is not worth the moat, there are multiple equivalent models, GLM 5.1 and Kimi K2.5 being the open ones, GPT 5.4 and Gemini 3.1 Pro being closed. https://llm-stats.com/ https://artificialanalysis.ai/leaderboards/models https://benchlm.ai/

Even API use (comparatively expensive) can be cheaper than Anthropic subscriptions if you properly use your agents to cache tokens, do context-heavy reading at the beginning of the session, and either keep prompt cache alive or cycle sessions frequently. Create tickets for subagents to do investigative work and use smaller cheaper models for that. Minimize your use of plugins, mcp, and skills.

Use cheaper models to do "non-intelligent" work (tool use, searching, writing docs/summaries) and expensive models for reasoning/problem-solving. Here's an example configuration: https://amirteymoori.com/opencode-multi-agent-setup-speciali... A more advanced one: https://vercel.com/kb/guide/how-i-use-opencode-with-vercel-a...

Re: Pro Max 5x quota exhausted in 1.5 hours despite moderate usage

#355

Post by bdangubic »

stavros wrote: post #343
comboy wrote: post #206

Earlier quoted context omitted.

You just convinced me to try it. Claude just copy pastes, does search and replace, zero abstractions and I'm the one that needs to think about the edge cases.

That's why I have Claude write the code and Codex review.

that’s like having oleg kiselyov’s code reviewed by my middle school daughter :)

Re: Pro Max 5x quota exhausted in 1.5 hours despite moderate usage

#356

Post by brunooliv »

stavros wrote: post #351
brunooliv wrote: post #321

Earlier quoted context omitted.

It's the nature of SaaS software, right? It doesn't need to be an enforced "hard change", but, let's say that they trained Opus 4.6 to be more "verbose" or to explore more files to gain more context for it's own tasks. If your limits stay "the same", but you then use Opus 4.6, your quota will be exhausted much faster, it's just how it works. Note that some features are simply NOT made for these Pro, Max, Max 5x or wh…

What you say makes sense, but they very actually reduced the token limits. We had, say, 20M tokens/week before, now we have 18M tokens/week (example numbers). They didn't just make a model that eats tokens faster.

Is that documented somewhere? Do you have a link? I might have just missed it, and if they did it, I will take my words back.

Re: Pro Max 5x quota exhausted in 1.5 hours despite moderate usage

#357

Post by fluidcruft »

throwdbaaway wrote: post #272
bcherny wrote: post #257

Earlier quoted context omitted.

> 1hr -> 5min on March 6th This is not accurate. The main agent typically uses a 1h cache (except for API customers, which can enable 1h but it is not on by default because it costs more). Sub-agents typically use a 5m cache.

https://github.com/anthropics/claude-code/issues/46829#issue... - Have you checked with your colleague? (and his AI, of course)

Doesn't what's said at the link approximately agree? The 5m bug was said to be isolated to use of overage (API billing).

Re: Pro Max 5x quota exhausted in 1.5 hours despite moderate usage

#358

Post by stavros »

bdangubic wrote: post #355
stavros wrote: post #343

Earlier quoted context omitted.

That's why I have Claude write the code and Codex review.

that’s like having oleg kiselyov’s code reviewed by my middle school daughter :)

I didn't know your middle school daughter is a genius coder, congratulations!

Re: Pro Max 5x quota exhausted in 1.5 hours despite moderate usage

#360

Post by toddmorey »

ozozozd wrote: post #336
toddmorey wrote: post #302

Earlier quoted context omitted.

My taxes are rather complex, so I ran the same exercise to see if Claude agreed with my accountant. An automated second opinion, so to speak. Spent about 6 minutes analyzing all the PDFs and basically nailed it perfectly in one shot. My only point here is it sure seems the same activity / use case can have wildly different results across sessions or users. Customer support and product development in the age of non-de…

What does nailing mean when you ask whether it agreed with your accountant?

Given the same inputs but not provided the results (output) from our accountant, did it come to the same conclusions or have good analysis as to why it differed?

Obviously, accounting is "spreadsheet math" intensive, so Claude wrote some python scripts for that which kept the math very stable. But there were some complex nuances that had taken the accountant and I quite a bit of work to track down and clarify. Claude quickly had a very accurate read on the situation and knew all the right clarifying questions.

I'm not yet ready to ever sign a return that's been entirely AI prepared, but I left the exercise pretty impressed.

Post reply on HN