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

GPT-5.4

openai.com

Post reply on HN
Print view View original

751–760 of 868 posts

Re: GPT-5.4

#751

Post by pmarreck »

Terretta wrote: post #734
pmarreck wrote: post #563

Earlier quoted context omitted.

so, since these companies have to comply with removing PII, is the worst thing that could happen to me, that I get ads that are more likely to be interesting to me? i’m not being facetious, honest question, especially considering ads are the only thing paying these people these days

Who has to comply with removing PII? Your profile, yours, mapped to a special snowflake ID, is packaged and sold across a network of 2500 - 4000 buyers, including in particular those that clean, tie (a surprisingly small footprint turns into its own "natural primary key"), qualify, and sell on to agencies. No step in this is illegal. https://www.theverge.com/2024/10/23/24277679/atlas-privacy-b...

my first and last name is already a "natural primary key" (every single google result of Peter Marreck is me), so I've already had to give that up a long time ago. So nothing new is lost I guess?

Re: GPT-5.4

#752

Post by sigbottle »

juanre wrote: post #748

I am running gpt-5.4 as one of my coding agents, and something interesting has happened: it's the first time I've seen an agent unfairly shift blame to a team mate: "Bob’s latest mail is actually the source of the confusion: he changed shared app/backend text to aweb/atlas. I’m correcting that with him now so we converge on the real model before any more code moves." This was very much not true; Eve (the agent writin…

Oh wow. I have noticed the GPT series was far more arrogant than its results showed sometimes (and unironically it digs in its heels even further when questioned on it). Opus rarely has this problem - but it goes a little too far in the opposite direction. Not totally sycophantic, but sometimes it can't differentiate genuine technical pushback because something is impossible, from suggestions or exploration.

Re: GPT-5.4

#753

Post by drik »

juanre wrote: post #748

I am running gpt-5.4 as one of my coding agents, and something interesting has happened: it's the first time I've seen an agent unfairly shift blame to a team mate: "Bob’s latest mail is actually the source of the confusion: he changed shared app/backend text to aweb/atlas. I’m correcting that with him now so we converge on the real model before any more code moves." This was very much not true; Eve (the agent writin…

how do you make them chat with each other?

Re: GPT-5.4

#754

Post by Razengan »

sigbottle wrote: post #752
juanre wrote: post #748

I am running gpt-5.4 as one of my coding agents, and something interesting has happened: it's the first time I've seen an agent unfairly shift blame to a team mate: "Bob’s latest mail is actually the source of the confusion: he changed shared app/backend text to aweb/atlas. I’m correcting that with him now so we converge on the real model before any more code moves." This was very much not true; Eve (the agent writin…

Oh wow. I have noticed the GPT series was far more arrogant than its results showed sometimes (and unironically it digs in its heels even further when questioned on it). Opus rarely has this problem - but it goes a little too far in the opposite direction. Not totally sycophantic, but sometimes it can't differentiate genuine technical pushback because something is impossible, from suggestions or exploration.

For me it's been the opposite. Are we getting A-B tested?

Re: GPT-5.4

#755

Post by Troniex-tech »

Looks more like context drift than “personality.”

When two agents coordinate, they’re mostly relying on compressed summaries of each other’s outputs. If one introduces a wrong assumption, the other often treats it as ground truth and builds on top of it. I’ve seen similar behavior in multi-agent coding loops where the model invents a causal explanation just to reconcile inconsistent state.

It’s that multi-agent setups need a stronger shared source of truth (repo diffs, state snapshots, etc.). Otherwise small context errors snowball fast.

Re: GPT-5.4

#756

Post by Razengan »

Philip-J-Fry wrote: post #374

I find it quite funny how this blog post has a big "Ask ChatGPT" box at the bottom. So you might think you could ask a question about the contents of the blog post, so you type the text "summarise this blog post". And it opens a new chat window with the link to the blog post followed by "summarise this blog post". Only to be told "I can't access external URLs directly, but if you can paste the relevant text or descri…

As bad as Google Gemini telling me it couldn't search Google Flights or Google reverse image search for me. These companies really need to dogfood their own products first. Do they not realize how embarrassing it is when their flagship intelligence refuses to interop with their own services?

Re: GPT-5.4

#757

Post by rishikeshs »

Philip-J-Fry wrote: post #374

I find it quite funny how this blog post has a big "Ask ChatGPT" box at the bottom. So you might think you could ask a question about the contents of the blog post, so you type the text "summarise this blog post". And it opens a new chat window with the link to the blog post followed by "summarise this blog post". Only to be told "I can't access external URLs directly, but if you can paste the relevant text or descri…

This is not only openai, but other models as well. Last week I added a summarise with AI block on a product blog page. I had seen it somewhere and felt like it’s a cool feature to have. Wrote a small shortcode in hugo for the block and added it with various models.

It’s like a hit and miss, sometimes claude says i cannot access your site which is not true.

Ref: https://formbeep.com/blog/building-formbeep-weekend/

Re: GPT-5.4

#758

Post by danesparza »

Razengan wrote: post #754
sigbottle wrote: post #752

Earlier quoted context omitted.

Oh wow. I have noticed the GPT series was far more arrogant than its results showed sometimes (and unironically it digs in its heels even further when questioned on it). Opus rarely has this problem - but it goes a little too far in the opposite direction. Not totally sycophantic, but sometimes it can't differentiate genuine technical pushback because something is impossible, from suggestions or exploration.

For me it's been the opposite. Are we getting A-B tested?

Yes.

Re: GPT-5.4

#759

Post by danesparza »

Razengan wrote: post #754
sigbottle wrote: post #752

Earlier quoted context omitted.

Oh wow. I have noticed the GPT series was far more arrogant than its results showed sometimes (and unironically it digs in its heels even further when questioned on it). Opus rarely has this problem - but it goes a little too far in the opposite direction. Not totally sycophantic, but sometimes it can't differentiate genuine technical pushback because something is impossible, from suggestions or exploration.

For me it's been the opposite. Are we getting A-B tested?

Or possibly: No

Re: GPT-5.4

#760

Post by Aurornis »

IgorPartola wrote: post #585
Aurornis wrote: post #387

Earlier quoted context omitted.

> What a model mess! OpenAI now has three price points: GPT 5.1, GPT 5.2 and now GPT 5.4. I don't know, this feels unnecessarily nitpicky to me It isn't hard to understand that 5.4 > 5.2 > 5.1. It's not hard to understand that the dash-variants have unique properties that you want to look up before selecting. Especially for a target audience of software engineers skipping a version number is a common occurrence and n…

The issue isn’t 5.4 > 5.2 etc. It is that there is a second dimension which is the model size and a third dimension which is what it is tuned for. And when you are releasing so quickly that flagship your instant mini model is on one numerical version but your flagship tool calling mini model is on another it is confusing trying to figure out which actual model you want for your use case. It’s not impossible to figure…

> The issue isn’t 5.4 > 5.2 etc. It is that there is a second dimension which is the model size and a third dimension which is what it is tuned for.

All 3 models are tuned for general purpose work.

Model size isn’t how you pick which model to use. You pick based on performance in evals compared to price.

It’s not hard to imagine that the more expensive models are probably larger or having higher compute requirements.

Post reply on HN