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

GPT-5.3-Codex

openai.com

Post reply on HN
Print view View original

581–590 of 634 posts

Re: GPT-5.3-Codex

#581

Post by vidarh »

rapind wrote: post #409
_zoltan_ wrote: post #297

Earlier quoted context omitted.

I'm personally 100% convinced of the opposite, that it's a waste of time to steer them. we know now that agentic loops can converge given the proper framing and self-reflectiveness tools.

Maybe some day, but as a claude code user it makes enough pretty serious screw ups, even with a very clearly defined plan, that I review everything it produces. You might be able to get away without the review step for a bit, but eventually (and not long) you will be bitten.

Reviewing what it produces once it thinks it has met the acceptance criteria and the test suite passes is very different from wasting time babysitting every tiny change.

Re: GPT-5.3-Codex

#582

Post by appsoftware »

timpera wrote: post #310
SunshineTheCat wrote: post #255

I've always been fascinated to see significantly more people talking about using Claude than I see people talking about Codex. I know that's anecdotal, but it just seems Claude is often the default. I'm sure there are key differences in how they handle coding tasks and maybe Claude is even a little better in some areas. However, the note I see the most from Claude users is running out of usage. Coding differences asi…

In my experience, OpenAI gives you unreasonable amounts of compute for €20/month. I am subscribed to both and Claude's limits are so tiny compared to ChatGPT's that it often feels like a rip-off. Claude also doesn't let you use a worse model after you reach your usage limits, which is a bit hard to swallow when you're paying for the service.

Claude when used via Github Co-Pilot is much better for useage allowance. I used Opus 4.5 for a months worth of development and only just hit 90 pct of the pro $40 per month allowance.

Re: GPT-5.3-Codex

#583

Post by PeterStuer »

xd1936 wrote: post #206
karmasimida wrote: post #198

Earlier quoted context omitted.

> With Codex (5.3), the framing is an interactive collaborator: you steer it mid-execution, stay in the loop, course-correct as it works. > With Opus 4.6, the emphasis is the opposite: a more autonomous, agentic, thoughtful system that plans deeply, runs longer, and asks less of the human. Ain't the UX is the exact opposite? Codex thinks much longer before gives you back the answer.

I've also had the exact opposite experience with tone. Claude Code wants to build with me, and Codex wants to go off on its own for a while before returning with opinions.

In terms of 'tone', I have been very impressed with Qwen-code-next over the last 2 days, especially as I have it running locally on a single modest 4090.

Re: GPT-5.3-Codex

#584

Post by anthonypasq »

ReptileMan wrote: post #286
RivieraKid wrote: post #284

Do software engineers here feel threatened by this? I certainly am. I'm surprised that this topic is almost entirely missing in these threads.

Jevons paradox hints that the situation is not as bleak as it sounds.

i think i fundamentally agree that the demand for code is essentially infinite. Code has just been notoriously expensive and therefore it could only de deployed towards the most economically efficient activities. this is now changing.

Re: GPT-5.3-Codex

#585

Post by AstroBen »

therealdrag0 wrote: post #492
AstroBen wrote: post #438

Earlier quoted context omitted.

What would things look like to make someone with currently ~10 years of experience unemployable? It's possible the job might change drastically, but I'm struggling to think of any scenario that doesn't also put most white collar professions out of work alongside me, and I don't think that's worth worrying about

Unemployable is a charged word. A lot of outdated professions still have professionals. There are still professional horse drawn carriages. If the AI performance gains are 50% improvement, and companies decide they rather cut costs and pocket the difference, could be due to many factors, that leaves millions out of a job. And those performance gains are coming for many white collar jobs. I guess your premise is mass…

> I guess your premise is mass unemployment is not worth worrying about, so okay then

My point wasn't that it's not a big deal. My point there is that if AI ends up taking a large % of white collar work you're going to have a huge portion of the population in the same boat. Maybe an overly optimistic view but that'll end up forcing change through politics

..I also think this is a ridiculously low % chance of happening and it would take something close to AGI to bring about. I don't know how you can use AI regularly and think we're anywhere close to that

Contracting an incurable illness that renders me blind and thus unable to work is just as likely and not something I spend time worrying about

> Marginal changes in productivity can make huge impacts to industries employment rates

Maybe? We also have Jevon's paradox. Software is incredibly expensive to build right now - how many more applications for it can people find if the cost halves?

Re: GPT-5.3-Codex

#586

Post by DrBazza »

It took several decades for the language server protocol and debugger server protocol (or whatever it's called). Is there a common 'agent' protocol yet? Or are these companies still in the walled-garden phase?

Re: GPT-5.3-Codex

#587

Post by adrianN »

freakynit wrote: post #517
adw wrote: post #458

Earlier quoted context omitted.

The prompt is decreasingly relevant. The verification environment you have is what actually matters.

I think this all comes down to information. Most prompts we give are severely information-deficient. The reason LLMs can still produce acceptable results is because they compensate with their prior training and background knowledge. The same applies to verification: it's fundamentally an information problem. You see this exact dynamic when delegating work to humans. That's why good teams rely on extremely detailed sp…

Having prompts be information deficient is the whole point of LLMs. The only complete description of a typical programming problem is the final code or an equivalent formal specification.

Re: GPT-5.3-Codex

#588

Post by arcanemachiner »

DrBazza wrote: post #586

It took several decades for the language server protocol and debugger server protocol (or whatever it's called). Is there a common 'agent' protocol yet? Or are these companies still in the walled-garden phase?

https://agentclientprotocol.com/get-started/introduction

Re: GPT-5.3-Codex

#590

Post by apetresc »

OutOfHere wrote: post #566
apetresc wrote: post #48

Earlier quoted context omitted.

Why is it absurd?

At the very least, it is absurd to announce a model but not release it on the same day, making it vaporware. Was it released?

Which model, 5.3 or 5.3-Codex? Yes, 5.3-Codex was announced and released. 5.3 wasn't announced. None of it is "absurd", and it also wouldn't have been "absurd" if they announce something but don't release it that same day (which they didn't do, but if they had - what exactly is absurd about that? Companies make announcements about future releases ALL the time.)
Post reply on HN