131–140 of 695 posts
Earlier quoted context omitted.
I had a weird experience at work last week where Claude was just thinking forever about tasks and not actually doing anything. It was unusable. The next day it was fine again.
This happened to me as well! It was especially infuriating because I had just barely upgraded to the $200 per month plan because I exhausted my weekly quota. Then the entire next day was a complete bust because of this issue. I want my money back!
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.
I spend full 20x the week quota in less than 10 hours. How is that possible? Well try to mass translate texts in 30 languages and you will hit limits extremely quick.
That's a really gnarly task but I'm shocked it burns 20x that fast. How large is the text? That matters more than anything.
It's a bit shocking to me how opaque the pricing for the subscription services by the frontier labs is. It's basically impossible for people to tell what they're actually buying, and difficult to even meaningfully report or compare experiences. How is this normal?
Yeah, I cancelled the moment I realized that the subscription is a scheme to get you to constantly dip into extra usage. I get more benefit out of Claude on the free tier than on Pro.
But the opacity itself is a bit offensive to me. It feels shady somehow.
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…
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…
Think Twitter's fail-whale problems. Sometimes you are lucky, sometimes you aren't. Why? We won't know until Anthropic figures it out and from the outside it sure looks like they're struggling.
Claude has gotten noticeably worse for me too. It goes into long exploration loops for 5+ minutes even when I point it to the exact files to inspect. Then 30 minutes later I hit session limits. Three sessions like that in a day, and suddenly 25% of the weekly limit is gone. I ended up buying the $100 Codex plan. So far it has been much more generous with usage and more accurate than Claude for the kind of work I do.…
This is what I'm working on proving now.
It is more that there is a confidence score while thinking. Opus will quit if it is too high and will grind on if the confidence score is close to the real answer. Haiku handles this well too.
If you give Sonnet a hard task, it won't quit when it should.
Nonetheless, that issue has been fixed with Opus.
I'll try to show that the speed of using Opus on tasks that have medium to hard difficultly is consistently the same price or cheaper than running them with Haiku and Sonnet. While easier tasks, the busy work that is known, is cheaper run with Haiku.
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…
As of now, I’m consistently hitting my 5 hour limit in less than 1 hour during N/A business hours. I’m getting to the point where I basically can’t use CC for work unless I work very early or late in the day.
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…
However, I've found that the flash quota is much more generous. I have been building a trio drive FOC system for the STM32G474 and basically prompting my way through the process. I have yet to be able to run completely out of flash quota in a given five hour time window. It is definitely completing the work a lot faster than I could do myself -- mainly due to its patience with trying different things to get to the bottom of problems. It's not perfect but it's pretty good. You do often have to pop back in and clean up debris left from debugging or attempts that went nowhere, or prompt the AI to do so, but that's a lot easier than figuring things out in the first place as long as you keep up with it.
I say this as someone who was really skeptical of AI coding until fairly recently. A friend gave me a tutorial last weekend, basically pointing out that you need to instruct the AI to test everything. Getting hardware-in-loop unit tests up and running was a big turning point for productivity on this project. I also self-wired a bunch of the peripherals on my dev board so that the unit tests could pretend to be connected to real external devices.
I think it helps a lot that I've been programming for the last twenty years, so I can sometimes jump in when it looks like the AI is spinning its wheels. But anyway, that's my experience. I'm just using flash and plan mode for everything and not running out of the $20/mo quota, probably getting things done 3x as fast as I could if I were writing everything myself.
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…
At least up until recently the 1M model was separated into /model opus[1M]