Despite the fact that their models are used in hiring, business, education, etc this multibillion company uses one benchmark with very artificial questions (BBQ) to evaluate how fair their model is. I am a little bit disappointed.
It's because these industries don't create their own benchmarks. The only ones creating evals are the AI company themselves or open source software engineers
I actually think there's a high chance that this curve becomes almost vertical at some point around a few hours. I think in less than 1 hour regime, scaling the time scales the complexity which the agent must internalize. While after a few hours, limitations of humans means we have to divide into subtasks/abstractions each of which are bounded in complexity which must be internalized. And there's a separate category of skills which are needed like abstraction, subgoal creation, error correction. It's a flimsy argument but I don't see scaling time of tasks for humans as a very reliable metric at all.
"GPT-5 in the API is simpler: it’s available as three models—regular, mini and nano—which can each be run at one of four reasoning levels: minimal (a new level not previously available for other OpenAI reasoning models), low, medium or high."
Is it actually simpler? For those who are currently using GPT 4.1, we're going from 3 options (4.1, 4.1 mini and 4.1 nano) to at least 8, if we don't consider gpt 5 regular - we now will have to choose between gpt 5 mini minimal, gpt 5 mini low, gpt 5 mini medium, gpt 5 mini high, gpt 5 nano minimal, gpt 5 nano low, gpt 5 nano medium and gpt 5 nano high.
And, while choosing between all these options, we'll always have to wonder: should I try adjusting the prompt that I'm using, or simply change the gpt 5 version or its reasoning level?
> but for the moment here’s the pelican I got from GPT-5 running at its default “medium” reasoning effort:
Would been interesting to see a comparison between low, medium and high reasoning_effort pelicans :)
When I've played around with GPT-OSS-120b recently, seems the difference in the final answer is huge, where "low" is essentially "no reasoning" and with "high" it can spend seemingly endless amount of tokens. I'm guessing the difference with GPT-5 will be similar?
Can anyone explain to me why they've removed parameter controls for temperature and top-p in reasoning models, including gpt-5? It strikes me that it makes it harder to build with these to do small tasks requiring high-levels of consistency, and in the API, I really value the ability to set certain tasks to a low temp.
It's cool and I'm glad it sounds like it's getting more reliable, but given the types of things people have been saying GPT-5 would be for the last two years you'd expect GPT-5 to be a world-shattering release rather than incremental and stable improvement.
It does sort of give me the vibe that the pure scaling maximalism really is dying off though. If the approach is on writing better routers, tooling, comboing specialized submodels on tasks, then it feels like there's a search for new ways to improve performance(and lower cost), suggesting the other established approaches weren't working. I could totally be wrong, but I feel like if just throwing more compute at the problem was working OpenAI probably wouldn't be spending much time on optimizing the user routing on currently existing strategies to get marginal improvements on average user interactions.
I've been pretty negative on the thesis of only needing more data/compute to achieve AGI with current techniques though, so perhaps I'm overly biased against it. If there's one thing that bothers me in general about the situation though, it's that it feels like we really have no clue what the actual status of these models is because of how closed off all the industry labs have become + the feeling of not being able to expect anything other than marketing language from the presentations. I suppose that's inevitable with the massive investments though. Maybe they've got some massive earthshattering model release coming out next, who knows.
> It does sort of give me the vibe that the pure scaling maximalism really is dying off though
I think the big question is if/when investors will start giving money to those who have been predicting this (with evidence) and trying other avenues.
Really though, why put all your eggs in one basket? That's what I've been confused about for awhile. Why fund yet another LLMs to AGI startup. Space is saturated with big players and has been for years. Even if LLMs could get there that doesn't mean something else won't get there faster and for less. It also seems you'd want a backup in order to avoid popping the bubble. Technology S-Curves and all that still apply to AI
Though I'm similarly biased, but so is everyone I know with a strong math and/or science background (I even mentioned it in my thesis more than a few times lol). Scaling is all you need just doesn't check out
>you'd expect GPT-5 to be a world-shattering release rather than incremental and stable improvement.
Compared to the GPT-4 release which was a little over 2 years ago (less than the gap between 3 and 4), it is. The only difference is we now have multiple organizations releasing state of the art models every few months. Even if models are improving at the same rate, those same big jumps after every handful of months was never realistic.
It's an incremental stable improvement over o3, which was released what? 4 months ago.
My reading is more that unit economics are starting to catch up with the frontier labs, rather than "scaling maximalism is dying". Maybe that is the same thing.
Isn’t reasoning, aka test-time compute, ultimately just another form of scaling? Yes it happens at a different stage, but the equation is still 'scale total compute > more intelligence'. In that sense, combining their biggest pre-trained models with their best reasoning strategies from RL could be the most impactful scaling lever available to them at the moment.
It seems like few people are referencing the improvements in reliability and deception. If the benchmarks given generalize, what OpenAI has in GPT-5 is a cheap, powerful, _reliable_ model -- the perfect engine to generate high quality synthetic data to punch through the training data bottleneck.
I'd expect that at some level of reliability this could lead to a self-improvement cycle, similar to how a powerful enough model (the Claude 4 models in Claude Code) enables iteratively converging on a solution to a problem even if it can't one-shot it.
No idea if we're at that point yet, but it seems a natural use for a model with these characteristics.
I mostly use Gemini 2.5 Pro. I have a “you are my editor” prompt asking it to proofread my texts. Recently it pointed out two typos in two different words that just weren’t there. Indeed, the two words each had a typo but not the one pointed out by Gemini.
The real typos were random missing letters. But the typos Gemini hallucinated were ones that are very common typos made in those words.
The only thing transformer based LLMs can ever do is _faking_ intelligence.
Which for many tasks is good enough. Even in my example above, the corrected text was flawless.
But for a whole category of tasks, LLMs without oversight will never be good enough because there simply is no real intelligence in them.
It reminds me of the latest, most advanced steam locomotives from the beginning of the 20th century.
They become extremely complex and sophisticated machines to squeeze a few more percent of efficiency compared to earlier models. Then diesel, and eventually electric locomotive arrived, much better and also much simpler than those late steam monsters.
I feel like that's where we are with LLM: extremely smart engineering to marginally improve quality, while increasing cost and complexity greatly. At some point we'll need a different approach if we want a world-shattering release.
Year on year, progress is indeed a bit incremental. But seen over a five year period the progress is actually stupidly amazing.
In practical terms, Gpt 5 is a nice upgrade over most other models. We'll no doubt get lots of subjective reports how it was wrong or right or worse than some other model for some chats. But my personal (subjective) experience so far is that it just made it possible for me to use codex on more serious projects. It still gets plenty of things wrong. But that's more because of a lack of context than hallucination issues. Context fixes are a lot easier than model improvements. But last week I didn't bother and now I'm getting decent results.
I don't really care what version number they slap on things. That is indeed just marketing. And competition is quite fierce so I can understand why they are overselling what could have been just chat gpt 4.2 or whatever.
Also discussions about AGI tend to bore me as they seem to escalate into low quality philosophical debates with lots of amateurs rehashing ancient argument poorly. There aren't a hell of a lot new arguments that people come up with at this point.
IMHO we don't actually need an AGI to bootstrap the singularity. We just AIs to be good enough to come up with algorithmic optimizations, breakthroughs and improvements at a steady pace. We're getting quite close to that and I wouldn't be surprised to learn that OpenAI's people are already eating their own dogfood in liberal quantities. It's not necessary for AIs to be conscious in order to come up with the improvements that might eventually enable such a thing. I expect the singularity might be more of a phase than a moment. And if you know your boiling frog analogy, we might be smack down in the middle of that already and just not realize it.
Five years ago, it was all very theoretical. And now I'm waiting for codex to wrap up a few pull requests that would have distracted me for a week each five years ago. It's taking too long and I'm procrastinating my gained productivity away on HN. But what else is new ;-).
Without that option the API will often provide a lengthy delay while the model burns through thinking tokens until you start getting back visible tokens for the final response.
> a real-time router that quickly decides which model to use based on conversation type, complexity, tool needs, and explicit intent
This is sort of interesting to me. It strikes me that so far we've had more or less direct access to the underlying model (apart from the system prompt and guardrails), but I wonder if going forward there's going to be more and more infrastructure between us and the model.
The router seems to only apply to the ChatGPT version, not the API, so it’s not really anything new. Gemini already has functionally dynamic reasoning effort.
"They claim impressive reductions in hallucinations. In my own usage I’ve not spotted a single hallucination yet, but that’s been true for me for Claude 4 and o3 recently as well—hallucination is so much less of a problem with this year’s models."
This has me so confused, Claude 4 (Sonnet and Opus) hallucinates daily for me, on both simple and hard things. And this is for small isolated questions at that.
There were also several hallucinations during the announcement. (I also see hallucinations every time I use Claude and GPT, which is several times a week. Paid and free tiers)
So not seeing them means either lying or incompetent. I always try to attribute to stupidity rather than malice (Hanlon's razor).
The big problem of LLMs is that they optimize human preference. This means they optimize for hidden errors.
Personally I'm really cautious about using tools that have stealthy failure modes. They just lead to many problems and lots of wasted hours debugging, even when failure rates are low. It just causes everything to slow down for me as I'm double checking everything and need to be much more meticulous if I know it's hard to see. It's like having a line of Python indented with an inconsistent white space character. Impossible to see. But what if you didn't have the interpreter telling you which line you failed on or being able to search or highlight these different characters. At least in this case you'd know there's an error. It's hard enough dealing with human generated invisible errors, but this just seems to perpetuate the LGTM crowd
"Are you GPT5" - No I'm 4o, 5 hasnt been released yet. "It was released today". Oh you're right, Im GPT5. You have reached the limit of the free usage of 4o
I'm curious what platform people are using to test GPT-5? I'm so deep into the claude code world that I'm actually unsure what the best option is outside of claude code...
It seems to be trained to use tools effectively to gather context. In this example against 4.1 and o3 it used 6 in the first turn in a pretty cool way (fetching different categories that could be relevant). Token use increases with that kind of tool calling but the aggressive pricing should make that moot. You could probably get it to not be so tool happy with prompting as well.
52 comments
[ 2.6 ms ] story [ 76.1 ms ] thread> and minimizing sycophancy
Now we're talking about a good feature! Actually one of my biggest annoyances with Cursor (that mostly uses Sonnet).
"You're absolutely right!"
I mean not really Cursor, but ok. I'll be super excited if we can get rid of these sycophancy tokens.
Is it actually simpler? For those who are currently using GPT 4.1, we're going from 3 options (4.1, 4.1 mini and 4.1 nano) to at least 8, if we don't consider gpt 5 regular - we now will have to choose between gpt 5 mini minimal, gpt 5 mini low, gpt 5 mini medium, gpt 5 mini high, gpt 5 nano minimal, gpt 5 nano low, gpt 5 nano medium and gpt 5 nano high.
And, while choosing between all these options, we'll always have to wonder: should I try adjusting the prompt that I'm using, or simply change the gpt 5 version or its reasoning level?
Would been interesting to see a comparison between low, medium and high reasoning_effort pelicans :)
When I've played around with GPT-OSS-120b recently, seems the difference in the final answer is huge, where "low" is essentially "no reasoning" and with "high" it can spend seemingly endless amount of tokens. I'm guessing the difference with GPT-5 will be similar?
Their PRO models were not (IMHO) worth 10X that of PLUS!
Not even close.
Especially when new competitors (eg. z.ai) are offering very compelling competition.
right :-D
It does sort of give me the vibe that the pure scaling maximalism really is dying off though. If the approach is on writing better routers, tooling, comboing specialized submodels on tasks, then it feels like there's a search for new ways to improve performance(and lower cost), suggesting the other established approaches weren't working. I could totally be wrong, but I feel like if just throwing more compute at the problem was working OpenAI probably wouldn't be spending much time on optimizing the user routing on currently existing strategies to get marginal improvements on average user interactions.
I've been pretty negative on the thesis of only needing more data/compute to achieve AGI with current techniques though, so perhaps I'm overly biased against it. If there's one thing that bothers me in general about the situation though, it's that it feels like we really have no clue what the actual status of these models is because of how closed off all the industry labs have become + the feeling of not being able to expect anything other than marketing language from the presentations. I suppose that's inevitable with the massive investments though. Maybe they've got some massive earthshattering model release coming out next, who knows.
Really though, why put all your eggs in one basket? That's what I've been confused about for awhile. Why fund yet another LLMs to AGI startup. Space is saturated with big players and has been for years. Even if LLMs could get there that doesn't mean something else won't get there faster and for less. It also seems you'd want a backup in order to avoid popping the bubble. Technology S-Curves and all that still apply to AI
Though I'm similarly biased, but so is everyone I know with a strong math and/or science background (I even mentioned it in my thesis more than a few times lol). Scaling is all you need just doesn't check out
Compared to the GPT-4 release which was a little over 2 years ago (less than the gap between 3 and 4), it is. The only difference is we now have multiple organizations releasing state of the art models every few months. Even if models are improving at the same rate, those same big jumps after every handful of months was never realistic.
It's an incremental stable improvement over o3, which was released what? 4 months ago.
I'd expect that at some level of reliability this could lead to a self-improvement cycle, similar to how a powerful enough model (the Claude 4 models in Claude Code) enables iteratively converging on a solution to a problem even if it can't one-shot it.
No idea if we're at that point yet, but it seems a natural use for a model with these characteristics.
The real typos were random missing letters. But the typos Gemini hallucinated were ones that are very common typos made in those words.
The only thing transformer based LLMs can ever do is _faking_ intelligence.
Which for many tasks is good enough. Even in my example above, the corrected text was flawless.
But for a whole category of tasks, LLMs without oversight will never be good enough because there simply is no real intelligence in them.
They become extremely complex and sophisticated machines to squeeze a few more percent of efficiency compared to earlier models. Then diesel, and eventually electric locomotive arrived, much better and also much simpler than those late steam monsters.
I feel like that's where we are with LLM: extremely smart engineering to marginally improve quality, while increasing cost and complexity greatly. At some point we'll need a different approach if we want a world-shattering release.
In practical terms, Gpt 5 is a nice upgrade over most other models. We'll no doubt get lots of subjective reports how it was wrong or right or worse than some other model for some chats. But my personal (subjective) experience so far is that it just made it possible for me to use codex on more serious projects. It still gets plenty of things wrong. But that's more because of a lack of context than hallucination issues. Context fixes are a lot easier than model improvements. But last week I didn't bother and now I'm getting decent results.
I don't really care what version number they slap on things. That is indeed just marketing. And competition is quite fierce so I can understand why they are overselling what could have been just chat gpt 4.2 or whatever.
Also discussions about AGI tend to bore me as they seem to escalate into low quality philosophical debates with lots of amateurs rehashing ancient argument poorly. There aren't a hell of a lot new arguments that people come up with at this point.
IMHO we don't actually need an AGI to bootstrap the singularity. We just AIs to be good enough to come up with algorithmic optimizations, breakthroughs and improvements at a steady pace. We're getting quite close to that and I wouldn't be surprised to learn that OpenAI's people are already eating their own dogfood in liberal quantities. It's not necessary for AIs to be conscious in order to come up with the improvements that might eventually enable such a thing. I expect the singularity might be more of a phase than a moment. And if you know your boiling frog analogy, we might be smack down in the middle of that already and just not realize it.
Five years ago, it was all very theoretical. And now I'm waiting for codex to wrap up a few pull requests that would have distracted me for a week each five years ago. It's taking too long and I'm procrastinating my gained productivity away on HN. But what else is new ;-).
> -------------------------------
"reasoning": {"summary": "auto"} }'
Here’s the response from that API call.
https://gist.github.com/simonw/1d1013ba059af76461153722005a0...
Without that option the API will often provide a lengthy delay while the model burns through thinking tokens until you start getting back visible tokens for the final response.
This is sort of interesting to me. It strikes me that so far we've had more or less direct access to the underlying model (apart from the system prompt and guardrails), but I wonder if going forward there's going to be more and more infrastructure between us and the model.
This has me so confused, Claude 4 (Sonnet and Opus) hallucinates daily for me, on both simple and hard things. And this is for small isolated questions at that.
So not seeing them means either lying or incompetent. I always try to attribute to stupidity rather than malice (Hanlon's razor).
The big problem of LLMs is that they optimize human preference. This means they optimize for hidden errors.
Personally I'm really cautious about using tools that have stealthy failure modes. They just lead to many problems and lots of wasted hours debugging, even when failure rates are low. It just causes everything to slow down for me as I'm double checking everything and need to be much more meticulous if I know it's hard to see. It's like having a line of Python indented with an inconsistent white space character. Impossible to see. But what if you didn't have the interpreter telling you which line you failed on or being able to search or highlight these different characters. At least in this case you'd know there's an error. It's hard enough dealing with human generated invisible errors, but this just seems to perpetuate the LGTM crowd
https://promptslice.com/share/b-2ap_rfjeJgIQsG
No mention about the (missing) elephant on the room, where are the benchmarks?
@simonw has been compromised. Sad.