[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / s / t / u / v / vg / vm / vmg / vr / vrpg / vst / w / wg] [i / ic] [r9k / s4s / vip] [cm / hm / lgbt / y] [3 / aco / adv / an / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / pw / qst / sci / soc / sp / tg / toy / trv / tv / vp / vt / wsg / wsr / x / xs] [Settings] [Search] [Mobile] [Home]
Board
Settings Mobile Home
/g/ - Technology


Thread archived.
You cannot reply anymore.


[Advertise on 4chan]


File: slop code.jpg (155 KB, 750x498)
155 KB JPG
Worth spending 20 bucks to try out Claude Code or Codex for a bit?
Got Google Antigravity through the free educational trial and while very convenient for quick prototyping and experimentation, it leaves a horrible trail of technical debt on whatever it touches and likes to break things that it was never asked to touch.
Is Claude or Codex any better?
>>
>>108954707
yes both are massively better than google slop, codex limits are much more generous than claude (especially when not using 5.5 and very high reasoning for basic tasks), but it's been getting worse too.
>>
>>108954707
OpenCode is better
>>
They all do this.
Use case for caring about technical debt? It just werks bro, who cares. Just get a better computer if it starts running slow.
>>
>>108954707
technical debt is usually the pebcak issue by not keeping the agent in check
>>
>>108954707
>just to be clear. i want you to clean up the abstractions as well. there are 36 tracked files in git and I think we can bring this number down primarily because we can now model sql lite in the shared project so we don't need to abstract specific implementations away. so i think we can bring that 36 number down to something more manageable to reason about
>>
>>108954721
Went with Codex for now, curious to see how it will fare.
>>
I've been using a Cursor subscription for the last month and it's been pretty good - It's built-in composer model does just fine on basic tasks and $20 of Claude API access bundled with it can be used to target trickier tasks. I've got a month of casual vibe coding out of it using that approach.

Would a Claude sub give me the same or better result? I'm hesitant because using the API chewed through credit really fast, like 10 requests across the month was enough. I don't want to subscribe and find I'm out of juice after a few hours....
>>
File: 1266250757001.png (5 KB, 400x400)
5 KB PNG
>>108957671
>give it fairly complex, but managable task
>immediatly blows through the 5 hour token limit in 30 minutes and 2 prompts (most of the time spent building)
Code quality seems quite good so far, but what in the hell are these limits?
>>
>>108954707
>20 bucks to try out Claude Code or Codex for a bit?
enjoy your two minutes of usage
>>
>>108955413
opencode is an eternal beta with multiple breaking updates every day
>>
>>108958132
generating new code is expensive. if you're with codex you should be able to see how many 'credits' you've used. 1000 credits is 40$.
It's going to get a lot worse soon because the subscription is way more generous than the api rates : just today I've apparently used ~67 000 credits.
>>
>>108958294
Seems like human programmers are starting to get cheaper than AI again.
>>
>>108954707
>Worth spending 20 bucks to try out Claude Code
if you want to watch it do half a plan maybe. doing anything start-to-finish requires either weeks of edging usage limits or spending a few hundred dollars in tokens.

I don't mind spending $200-300 every couple of months when I require a working MVP in a stack I am not familiar with YESTERDAY but you need to be realistic about these things.



[Advertise on 4chan]

Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.