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

171–180 of 695 posts

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

#171

Post by ryandrake »

meetingthrower wrote: post #56

I don't get it. Last week on the 100 bucks plan I generated probably 50k LOC (not a quality measure for sure!) and just barely kissed the weekly limit. I did get rate limited on some sessions for sure, but that's to be expected. I'm curious what are people doing that is consuming your limits? I can't imagine filling the $200 a month plan unless I was essentially using Claude code itself as the api to mass process stu…

Yea, I found myself maxing out the $20/mo plan occasionally, so I tried the $100/mo, but I don't think I even once even approached the session limit, let alone the weekly limit. And this is doing what I would consider heavy, continuous programming. I probably ought to go back down to $20 one. It would be nice if they had a cheaper tier in between them, but the tiers they have are probably a good business trick to get people to buy much more than they need.

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

#172

Post by egeozcan »

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.

This exact thing is happening to me since yesterday. It comes back to life when I throw the whole session away.

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

#173

Post by himata4113 »

SkyPuncher wrote: post #125

I skimmed the issue. No wonder Anthropic closes these tickets out without much action. That’s just a wall of AI garbage. Here’s what I’ve done to mostly fix my usage issues: * Turn on max thinking on every session. It save tokens overall because I’m not correcting it of having it waste energy on dead paths. * keep active sessions active. It seems like caches are expiring after ~5 minutes (especially during peak usage…

The problem is actually because their cache invalidates randomly so that's why replaying inputs at 200k+ and above sucks up all usage. This is a bug within their systems that they refuse to acknowledge. My guess is that API clients kick off subscription users cache early which explains this behavior, if so then it's a feature not a bug.

They also silently raised the usage input tokens consume so it's a double whammi.

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

#174

Post by ejpir »

scrollop wrote: post #132
jedisct1 wrote: post #7

GPT-5.4 works amazingly well. I’ve moved away from Claude and toward open-source models plus a ChatGPT subscription. That setup has worked really well for me: the subscription is generous, the API is flexible, and it fits nicely into my workflow. GPT-5.4 + Swival ( https://swival.dev ) are now my daily drivers.

Chatgpt has better limits however the responses even on 5.4 xtra thinking are not as good as sonnets. Wish Claude would get their house in order.

Show us some reciepts in the form of a exported session. I've been a heavy user of Claude up untill the end of feb, but switched to Codex because it's better at handling large code bases, following the "plan", implementing the backend changes in Zig. If you ask Claude to do a review of the code and suggest fixes, then let it Codex review it, then again ask Claude, it will 99% of the time say. Oh yes you are right, let me fix that.

Either you are using it wrong or you are working in a totally different field.

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

#175

Post by jerf »

1970-01-01 wrote: post #74
geeky4qwerty wrote: post #39

I'm afraid the music may be slowly fading at this party, and the lights will soon be turned on. We may very well look back on the last couple years as the golden era of subsidized GenAI compute. For those not in the Google Gemini/Antigravity sphere, over the last month or so that community has been experiencing nothing short of contempt from Google when attempting to address an apparent bait and switch on quota expec…

Lights on = Ads in your output. EOY latest; they can't keep kicking the massive costs down the road.

Ads do not pay enough to cover AI usage. People see the big numbers Google and Facebook make in ads and forget to divide the number by the number of people they serve ads to, let alone the number of ads they served to get to that per-user number. You can't pay for 3 cents of inference with .07 cents of revenue.

You also can't put ads in code completion AIs because the instant you do the utility to me of them at work drops to negative. Guess how much money companies are going to pay for negative-value AIs? Let's just say it won't exactly pay for the AI bubble. A code agent AI puts an ad for, well, anything and the AI accidentally puts it into code that gets served out to a customer and someone's going to sue. The merits of the case won't matter, nor the fact the customer "should have caught it in review", the lawsuit and public reputation hit (how many people here are reading this and salivating at the thought of being able to post an angrygram about AIs being nothing but ad machines?) still cost way too much for the AI companies creating the agents to risk.

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

#177

Post by throwaway2027 »

weavie wrote: post #90

How good are local LLMs at coding these days? Does anyone have any recommendations for how to get this setup? What would the minimum spend be for usable hardware? I am getting bored of having to plan my weekends around quota limit reset times...

Some claim that some of the recent smaller local models are as good as Sonnet 4.5 of last year and the bigger high-end models can be as almost as good as Claude, Gemini and Codex today, but some say they're benchmaxed and not representative.

To try things out you can use llama.cpp with Vulkan or even CPU and a small model like Gemma 4 26B-A4B or Gemma 4 31B or Qwen 3.5 35-A3B or Qwen3.5 27B. Some of the smaller quants fit within 16GB of GPU memory. The default people usually go with now is Q4_K_XL, a 4-bit quant for decent performance and size.

https://huggingface.co/unsloth/gemma-4-26B-A4B-it-GGUF

https://huggingface.co/unsloth/gemma-4-31B-it-GGUF

https://huggingface.co/unsloth/Qwen3.5-35B-A3B-GGUF

https://huggingface.co/unsloth/Qwen3.5-27B-GGUF

Get a second hand 3090/4090 or buy a new Intel Arc Pro B70. Use MoE models and offload to RAM for best bang for your buck. For speed try to find a model that fits entirely within VRAM. If you want to use multiple GPUs you might want to switch to vLLM or something else.

You can try any of the following models:

High-end: GLM 5.1, MiniMax 2.7

Medium: Gemma 4, Qwen 3.5

https://unsloth.ai/docs/models/minimax-m27

https://unsloth.ai/docs/models/glm-5.1

https://unsloth.ai/docs/models/gemma-4

https://unsloth.ai/docs/models/qwen3.5

https://github.com/ggml-org/llama.cpp

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

#180

Post by iLoveOncall »

It's very easy to calculate the actual cost given they list the exact tokens used. If I take the AWS Bedrock pricing for Opus 4.6 1M context (because Anthropics APIs are subsidized and sold at a loss), here's what each costs:

Cache reads cost $0.31

Cache writes cost $105

Input tokens cost $0.04

Output tokens cost $28.75

The total spent in the session is $134.10, while the Pro Max 5x subscription is $100.

Even taking the Anthropics API pricing, we arrive at $80.58. Below the subscription price, but not by much.

It's just the end of the free tokens, nothing to see here. It's easy to feel like you're doing "moderate" or even "light" usage because you use so little input tokens, but those "agentic workflows" are simply not viable financially.

Post reply on HN