No; you're just more focused/aware of common outages (e.g. this, the bank "outage" that wasn't, etc) that occur every week that normally resolve themselves rapidly. The FAA one is the only legitimately exceptional outage (even Azure outages are fairly run-of-the-mill).
None of the recent popular outages have been the result of a cyber attack or AI, and nobody credible is claiming otherwise.
It hasn't stopped people repeating "AI or Cyber Attack did it!" to one another in every thread, and getting more and more sure because others believe it baselessly.
That's a ridiculous statement. Outlook down for a day is not a weekly event. FAA Ground Stops aren't a weekly event.
They are things that almost never happen.
While it might be a pattern of rushed upgrades, it's still a pattern that something isn't quite right. That might just be rushed upgrades due to fear of vulnerabilities getting exploited left and right, but it's still clearly a big difference from even 6 months ago.
- 3 times in 25 years, but the previous failure was less than 3 years ago. Individual airlines have had their own outages at least as severe as this one numerous times in recent years
- Outlook may have never had full feature failure before, but other Microsoft services have outages all the time. Azure is a dumpster fire. GitHub even more so. Remember when Crowdstrike bricked 9 million Windows computers less than 2 years ago?
- What scale? JPMorgan Chase had an issue for a few hours. Is that unprecedented?
- NSA outages are unlikely to be leaked in the first place. we have no idea how frequently something like this occurs.
Would people even be talking about any of these if they weren’t primed to be freaking out about it? In my circle the people freaking out the most are the people most removed from tech and software development, in no small part because they have no idea what constitutes normal and are reacting to headlines and social media feeds.
Then cite a source claiming the FAA, Outlook, or supposed bank outage, is linked to AI or a Cyber Attack. If you cannot link to a reliable source, it is by definition baseless.
The onus falls directly back to you to prove that these claims are unfounded. Prove that it's not a direct attack. Oh, and quoting sycophants and child rapists dont count.
The vendors themselves haven't stated such an "attack." Microsoft's released a detailed breakdown of what occurred and why on the Azure Admin portal which has been widely reposted:
So, I've cited my sources. It is now your turn to cite your "direct attack" citations. My claims have a basis, thus why I correctly called the claims of AI or Cyber Attacker baseless. That's what baseless means, as in without factual basis - which they are.
You've done nothing to explain why those are claims based in fact, and instead just deflected.
That page, at the time of writing seems to be focused around iCloud and or "iCloud Account & Sign In." Many of the services, that page claims, are up if you're already signed in.
That's my impression; a total of zero of the popular recent ones have been related to "AI" or Cyber attacks, but every comment in every thread, talks about those two as if they were the underlying cause.
The FAA's recent outage for example, at its core, seems like people trying to rush needed upgrades that were slow for a specific reason (e.g. reducing the testing window for 90 to 30-days).
The Microsoft outage seems like three different things linked together (and poor "full failure" recovery prep, lengthening the outage).
The bank "outage" may not have even existed, 2/3 of the banks claimed they didn't have an outage at all, and the other said it was a limited technical issue. The articles were based on DownDetector spikes.
You need to be more skeptical of what organizations say when they have a lot to lose. Nobody takes responsibility for situations unless there's way to offload some or all of the blame. They have every incentive to lie about the situation to make it go away. Politics often plays the biggest role in these situations.
Those things are usually integrated as some type of TPP and are not directly related to any core banking services: If Zelle connections fails, usually nothing gets broken in your "real" account data, since Zelle is connected by whatever API stack to the bank
It takes two to tango.
If $BANK_A has a problem transferring to $BANK_B does that necessarily mean both banks are experiencing an outage? Surely the common user (and therefore Downdetector) are likely to report a false-positive...
Are you some sort of psychic? While Microsoft has shared portions of its internal review with some journalists, and are supposedly planning on releasing it publicly next week, in the FAA, banks, and NSA cases we have very little information aside from non-technical statements from the affected with clear reasons not to immediately divulge verifiable details as their investigations get under way.
You're out here categorically saying none of these had an AI or cyber-attack angle with no verifiable proof and having a tantrum that other people are doing the same thing in reverse.
The list of unanswered questions is long. Maybe focus on pushing for more details instead of immediately saying "nothing to see here" regarding a series of important failures within a week's time.
BTW here's JPMorganChase's statement:
> JPMorganChase has experienced a temporary issue related to a software bug in on-premise hardware systems – there is no evidence this is cyber-related. Where we were party to a transaction involving other banks, those banks may have been impacted, but this is not a systemic issue. Customers still have access to their accounts and funds.
If you were trying to send your rent payment via Zelle from your Citibank account to your landlord's Chase account and it continually failed yesterday with no clear reason immediately provided, of course any complaint is a "witch hunt" since your Citibank account "works."
There is a near 100% chance they are sharing an interpretation that is more friendly to themselves than the facts, but a near 0 chance they are lying. A major outage at a large company or government department will have hundreds of people, at minimum, involved in the response. If the company lies about it, somebody will talk to the press. They know this, so they will not lie.
> You need to be more skeptical of what organizations say when they have a lot to lose.
So let me see if I understand what I should be:
- Be more skeptical of official explanations, including where it includes a technical breakdown of how it occurred (e.g. Microsoft/Outlook outage).
- Be less skeptical of random commenters claiming, without proof, that it is AI or Cyber Attacks.
Because that's the implied point you're making. Be more skeptical of A and less skeptical of B, even though A has someone placing their reputation behind it, and B is anonymous commenters throwing made up theories out into the ether.
> The FAA's recent outage for example, at its core, seems like people trying to rush needed upgrades that were slow for a specific reason (e.g. reducing the testing window for 90 to 30-days).
The explanation for this one still doesn't ring true for me and that's exacerbated by how many times the story has changed after telling the public that they didn't need to know the reasons.
Agree, though, that the rest do have reasonable explanations
IDK what the current explanation is but I remember even in the early 20 teens FAA workers were ringing alarms that the entire control system was critically out of date and in need of upgrades and repairs. That they were having issues sourcing reliable FLOPPY disks, which they STILL use in 2026. This of course went almost completely ignored.
Given how many companies seem to be working hard to have all programming work done by AI, and pushing out everyone who says this sounds like a bad idea, is "AI" really not going to be a significant part of "rushing out needed updates"?
Is "Claude says it looks good so let's skip the usual testing window and push to prod on a Friday" really not a sentence you can imagine being said in this current environment?
I am talking about outages that have already recently occurred. You're making a distinct point about a hypothetical.
My point is that people are taking a potential future problem and applying to actual things that have recently happened, that didn't fit that mold. Which is incorrect.
When AI or Cyber Attacks cause an outage, people are absolutely right to say so. But claiming - without proof, that every outage is related to the two is noise.
In my line of work I deal with a lot of banks. They have been absolutely panicked for the last few months as they have been using AI to find and exploit bugs in their own software (A huge number of banks are members of project glasswing and other AI cyber testing with less limits than the public ones we get). This has forced them to drop almost everything else and immediately address updates in their systems, especially to a lot of legacy software that's sat around forever.
Just this many updates is going to cause instability in itself. Old systems that aren't touched can run forever without many problems, but when you can RCE said app the normal 3+ months of testing goes out the window. Now imagine this with tons of bugs in hundreds or thousands of applications at a single bank and you can see why things are starting to get unstable.
The article is confusing, they state the problem, then bring up two examples both of which broke even in terms of costs Vs. income, but then claimed local officials were disappointed they weren't profitable instead.
I thought the point of these things was to get people to pull off of major roads, into local towns/areas, and potentially spend money at local businesses. If they're breaking even, and still accomplishing that, then there isn't a problem?
Until the things wear out and need to be replaced. Many of them were built in older/worse safety standards - they have "worked", but they would not be allowed for good reason today and should be torn down.
I don't live in Paris, so I don't know what is best. I'm sure that some people who live there would prefer a Costco. I'm not catholic (like many people in France) so while I appreciate the architecture I otherwise have no particular use for that building - I'm more likely to visit it as a foreigner (as part of some tour) than if I was a local (where I could anytime, but in practice would likely always find something else to do).
The city is very far from the major highway and it has its own world-class big tourist attraction: the Royal Tyrrell Museum[1], i.m.o. by far the best dinosaur museum I've ever seen.
Yes, I've been to New York's, London's and Vienna's natural history museums. When it comes strictly to dinosaurs, Drumheller is still better. But on the remainder of natural history Vienna's is better than all others.
Because the exact same processes that create it faster, are also now generating large comprehensive test suites, and data-validation suites based around the actual business requirements and data/edge cases.
In essence, it is writing the code faster (great), but then also writing the code that was often previously never written due to budget/time limitations (comprehensive testing/validation).
So, how do you know the code is the highest quality ever? Because we have test result data that PROVES it to be the case. It isn't "trust but verify" it is "Don't trust. Verify."
This is my observation as well. I thought I was good at testing (I wanted to be good at testing), but the models are just better at it. Testing in particular rewards a tireless cataloging of the tiny and the tedious and in this respect LLMs are unmatched.
Single data point, my Sentry notifications across more than a half dozen client projects when from 2-4 a month to basically 0. That's while shipping 20x _more_ code.
Yeah but you don't test for stuff you don't know you have to test for. All tests can pass and internally its still done in a braindead way that will explode at some point.
Absolutely, and this is exactly why humans are weaker than LLMs. We consider what is likely, they consider what the data/logic shows is possible. So you get edge cases tests/flagged, that wouldn't even be something a real person would know to look for.
You must have replied to the wrong comment, since the one you did replied to never claimed that the test and validation suites were "never looked at." They are reviewed, and the test/data results are reviewed.
> Because we have test result data that PROVES it to be the case.
pgrust thought this too, but then another human discovered claude had hardcoded benchmark query plans.
Now pgrust has a "test" [1] to make sure the Ai doesn't do this again (unless it decides this test is bad for the current goal)
I have a growing list of these, another notable one is OpenCode having a security regression [2] because the Ai redesigned configuration loading for v2.
Real verification seems beyond transformer capabilities.
This is a roundabout way of saying you're leaving the profession.
We now have people commonly running multiple concurrent programming tasks, while automatically and interactively testing them, reviewing them, and validating them. They're then handed for human review, where little is found because the quality regime is now set so high. We're talking weeks of development + testing/validating time brought down into the realm of days.
So this is pure "my horse & cart are just as good as the motorcar"-levels of argument. If I bury my head in the sand long enough, and deep enough, this will all just blow over, and we'll go back to doing stuff significantly slower/worse any day now...
The quality of stuff I am shipping today is the highest it has ever been, and tested/validated significantly more comprehensively. Why? Because time/resources was the limitation, not willingness or skill. I always wanted to test for every single possible edge case/scenario, but it wasn't in the time-budget (and diminish returns). Now I don't need to make that trade-off.
And that's just the tip of the iceberg for the iOS Keyboard's issues. They've tried to fix it multiple times, and while it is undeniably better than its lowest point - it is still staggeringly worse than Google's native offering on the Pixel range. Swiping is just so painful, and Microsoft's Swiftkey keyboard is on life support.
Considering arguably the keyboard itself is one of the most common "touch surfaces" on iOS, the level of neglect and level of overall quality is staggering. There's some serious incentive problems at Apple if nobody is tasked with making one of the best digital keyboards in the world.
PS - Whoever came up with the concept, on swipe-style touch keyboards, of the "Tap Map"/Heatmap wherein keys visible location doesn't match their touch area - so that BY DESIGN it always provides an inconsistent user experience needs to be fired into the sun. You literally cannot disable it on most mobile keyboards. It is on iOS's native keyboard AND Microsoft's Swiftkey.
Of course; but my point is just add it to the giant pile of problems with iOS's keyboard. Every part of it has bugs, broken stuff, or annoyances like this, it isn't fit for purpose, and needs significant work.
The fact Apple aren't embarrassed by the iOS keyboard is very surprising and disappointing.
Every big company reaches a point where their business becomes selling their users out to other businesses.
Apple has been doing that for years:
- The iCloud Keychain API that lets Facebook, TikTok etc. track users across devices, app reinstalls and device resets.
- Ads in formerly ad-free built-in apps.
- Removing the Disable Javascript from Safari's menu so you can't easily disable the now-mandatory tracking bullshit on most sites.
- Ads
- "Force closing" apps on iOS not really force quitting them instantly anymore, giving each app a "last breath" chance to call home about anything it wants to send about you.
- Ads
- And now they've added nagging when you try to cancel recurring in-app subscriptions.
But removing the easy access button that breaks so many websites is not selling out users.
Giving an app some fraction of a second to clean up is not selling out users. This is a pretty ridiculous stretch on your part. If it had something to phone home it could have done it five seconds ago, why wait? Or it could probably do it the next time you open the app. That's not what the grace period is about.
Nobody was "accidentally" clicking that Disable Javascript button, what the hell
Oh and did you notice how for several macOS versions the "Manage Website Data…" button in Safari Settings, for selectively removing cookies, magically fails to work?
Oh OH and how iCloud+'s paid Private Relay feature doesn't work in authoritarian countries like the UAE etc where it would actually be needed?
Fully disabling javascript has become very niche. It's okay to not have that button and wanting a simpler UI with fewer footguns is plenty of justification to remove it from the menu. Even if I would occasionally use it myself.
It's more like removing a dumbwaiter. The reason to cut the button is that it's less and less useful and it's taking valuable space, not the mild risk. But the risk isn't helping.
Even lockdown mode doesn't disable javascript, just makes it slower.
(It's not as bad as a dumbwaiter but it's sure as heck nowhere near "stairs". What percent of people even know what that button does?)
It’s the other way around. Apple don’t want to suggest offensive phases and have it look like they wants you to say those things.
The middle ground here is to not suggest anything at all. Because they’re either pushing people to say something or preventing them from saying something.
The HN crowd forgets that average people don’t know how technology works. Everyone I know would see this as a good feature.
This should be a preference then, just like in video games where profanity can either be shown or hidden in multiplayer chats. Then it makes the user aware this exists as a feature and gives them control over it.
My rule of thumb in software is: if there's enough of a dispute on how something should work, it should probably turn into a preference available to users. It's adds scope, but it respects user choice in places where it matters.
Building software long enough I've realized that the valuable part is your vision of it. The opinionated part. There's no toggle to allow p0rn in the app store, and that's good. We can argue about user choice, but undeniably even if such a toggle existed, more users would fall into addiction. Environment matters more than "self control" or "discipline".
I agree with you (on certain types of software), and I also don't see Apple as the be-all and end-all or assume that its opinions are always right. This is just my opinion based on my lived experience, both as a user and as an engineer.
I still think the original iPhone keyboard in 2007 was better than what we have now. It’s been downhill ever since iOS 7.
I used to tell people just type and trust it to work, it was great. Now, even when I type the right thing, the auto-complete will screw it up after the fact. I feel like I need to proofread everything. I wish we could go back to whatever setup they had 19 years ago.
Autocorrect aside, the keyboard itself got broken around the iPhone 6 release. I've mostly had autocorrect disabled back then and now. It used to be so much better on the 5. And of course if you're horribly misspelling everything, the autocorrect will perform worse.
I remember it being whenever they first added the next-word predictions above the keyboard. I was guessing it was iOS 7, just because that was the massive UI overhaul, but it very well could have been iOS 6.
There are probably two unrelated things going on. Keyboard broke with iPhone 6 (which shipped with iOS 8), and autocorrect broke with iOS 6 or 7. I wouldn't know about the latter cause I had autocorrect disabled almost always.
I'd argue iPhone got ahead of the game then just stopped getting at all better. Siri AI is straight garbage, they won't let the interface be anything but what they want it to be, and the app store lock in/lack of ad blockers on browsers has made it basically just a phone for me. I have no intention of upgrading beyond my 12, and when it finally dies, I'm probably moving back to some kind of android phone if there's not a viable open source model out there by then (viable for everything, banking, etc). I only moved to apple because the watch was leagues better at the time than samsung's offering, but now I can't even claim that (and quit wearing it years ago)
In my experience it’s worse than it’s ever been in the 15 years I’ve been on iOS. It’s absolutely enraging to me. I’m too lazy and hate smart phones so much anyways that I am not investing mental cycles and emotion into switching. But it’s a disgrace how shitty it is.
While they both fall under DHS, you're confusing CBP with ICE.
CBP's job is to protect the border and stop drug and people trafficking entering, whereas ICE's original purpose was to enforce immigration law within the US.
While not "everything" ICE was tasked with was immigration, that was their bread and butter. Where the vast majority of their workforce and spend went.
But why is ICE, specifically, investigating that? The Federal government has multiple criminal divisions - why are ICE now a US Citizen criminal investigatory branch too? Seems far outside of their remit.
Almost all the cabinet level departments has their own Inspector General office since the 70s and 80s.
Homeland Security got one in 2003 with the combination of previous agencies.
The Homeland Security Act of 2002 also let those offices start having their own statutory powers and carry firearms without needing to be deputized by the US Marshals Service.
I work at OpenAI and I can try to explain our logic.
In general, LLM performance is incredibly multidimensional. There's no guarantee of a fixed ordering where the model that is best at math is also the model that's best at jokes. Despite that complexity, much of the variation in performance is explained by just a single variable: model size. In general, models that are bigger will (a) be more performant / reliable, (b) cost more per token, and (c) have better recall of esoteric facts. As you note, our models are ordered:
Luna < Terra < Sol < Astra
Even today, I'd recommend this as a general starting point for picking models. If you care about quality only, pick Astra. If you're a bit price sensitive, pick Sol. If you're very price sensitive, pick Terra or Luna.
Of course, model size doesn't capture all variation. Models that are trained later might have new data or new techniques or new architectures that earlier models lack. You can see this with models from all providers - Sonnet 5.5 outperforms Fable 5.1 in ways, and Gemini 3.8 Flash outperforms 3.1 Pro in ways.
In the case of GPT-6.1 Terra, I'd say Terra is probably better than Astra and Sol at technical communication, similar on vision and computer use (but much faster!), and a bit worse on most other things. For API developers, it's a nice midrange option if you want something cheaper than Sol but a lot smarter than Luna (especially for vision / computer use / communication). For subscribers, it's not a bombshell release - if you're hitting usage limits, maybe it helps you go a bit further, and if you want clearer communication or faster computer use, it can be a nice option to select.
If you have suggestions on how to name our products better, I'm all ears. Our intention is to make things clear and simple, even if we miss the mark sometimes.
Unfortunately, a lot of the complexity is somewhat irreducible and due to to the actual complexity of the products. Scaling model size is different from scaling reasoning effort which is different from scaling multi-agent count. So if we just swept all of those under a single name of "more IQ", it would make the naming simpler but the experience much worse, as important details would be obfuscated from end users.
Do you then recommend Sol for technical documentation?
What's confusing to me in addition to the naming is that there are no published guidelines I can use, as far as I can see. Put yourself in the user's shoes. On what basis should one pick a model for a given task?
I can understand smaller models being dumber and faster. That aligns well with scenarios where you are willing to compromise; e.g. you're classifying content or asking a model to format text into a table.
But when you want "maximum intelligence, cost and time be damned", there's the implication that bigger models are smarter on a linear scale, and every provider's model release naming is currently conflicting with that expectation.
This isn't just a naming problem, but a communication problem. If I knew that Astra was the best at coding but a bit bad at writing, and Sol was almost as smart but better at technical writing, then that's the kind of thing I expect to be documented.
One possible way: A simple matrix with horizontal columns like "writing", "coding", "math", rows are the models, and each box is a number of stars from 1-5.
Instead, this sort of thing becomes lore, shared by word of mouth.
> Do you then recommend Sol for technical documentation?
I don't know, to be honest. Technical documentation is such a multidimensional skill, and I also don't know your budget. I'd imagine Terra might have the most readable prose, but Astra might be more likely to catch bugs or mistakes to highlight. If the documentation is about a more esoteric language or situation, Astra might have a greater edge than if it's a common language or situation. If you have a long list of documentation guidelines, that might also give more of an edge to Astra. I wouldn't want to make a recommendation without doing a lot of testing, and I haven't done a lot of testing.
Some food for thought, if someone from oAI can't make a direct recommendation for something as direct as "Technical documentation is such a multidimensional skill" - how could we know assess what model to use?
To elaborate, I can ask each model to write a doc about something I know well and tell you which output I like better.
But to assess whether my preference on my problem is going to generalize across other people, other programming languages, other documentation types/formats, etc., and then analyze statistical significance is a thornier question that I don't have the data to answer. I'd rather make no recommendation than a weakly informed recommendation.
> I can ask each model to write a doc about something I know well and tell you which output I like better.
That's what users currently end up having to do on their own. Or they just give up in confusion, which presents in your site analytics as either bounce, buy later or buy less.
Why not have a model comparison site? Visitors choose from a dozen-ish broad categories of work (eg technical documentation, market communications, etc), then pick from a dozen sample prompts across typical tasks in that category, and see the model responses with avg time and approximate cost. Especially helpful would be having a toggle overlay with annotation highlighting relevant differences, then if they have (or create) an account, they can edit the annotations and add their own with a sharing link.
If my job was to actually help my co-workers and friends make better informed choices they feel more confident in, I'd be doing something like that. The extra credit version would help users narrow down to two models and then let them convert to a 'Test Drive' trial - same workflow but from their own prompts, side-by-side A/B response comparisons with annotation and sharing link.
As you said, the answer to "Which model is best for me?" is actually complicated, conditional and inevitably a bit confusing. When that's the reality, the right pattern isn't 'Communicate to Convert', it's 'Build credibility and confidence with maximum clarity and genuine helpfulness'.
People vote on pairs of model responses and they aggregate everyone's votes into a leaderboard.
One thing to be aware of that the distribution of prompts on arena.ai can be pretty different than real-world usage. The people who spend time on that website casting votes tend to be more interested in entertainment than professional productivity. And turn-based voting can punish useful behavior like asking clarifying questions instead of guessing at assumptions.
Yes, I've seen Arena.ai and the reasons you cited are why I suggested something better that's more private, personalized and shareable internally without being skewed and gamed.
It’s overkill most of the time, but frankly all these models are “cheap”, and I’d rather waste compute than brain cycles trying to discern which dials are best for each individual as-hoc query.
But Astra is simply bad at writing. I have to use Argon to rewrite everything Astra wrote in an academic setting. Astra always tries to cover every small detail and engages in defensive writing that is unreadable.
Totally agree. I'm working a lot faster when being just about everything I want on the first hop, without model switching and goofing around to find the sweet spot. When I need a janky answer I run a model locally with llama.cpp
> Put yourself in the user's shoes. On what basis should one pick a model for a given task?
Create a realistic eval suite for your use case using your actual data that you manually label, with easier, more complex, rarer samples, and run each model you are considering against it and decide based on the usual confusion matrix, cost, latency, or whatever metrics matter to you. There's really no other way to know – and you might be surprised that cheaper models can often perform your task(s) relatively well at a fraction of the cost.
> Create a realistic eval suite for your use case using your actual data that you manually label, with easier, more complex, rarer samples, and run each model you are considering against it and decide based on the usual confusion matrix, cost, latency, or whatever metrics matter to you.
I’d imagine there’s only a tiny percentage of users that can or want to do this.
if you consider every single person that uses some LLM then sure, it is tiny percentage, but I would wage that large percentage don't write technical documentation as GP. large userbase probably don't need THE best solution for their use case and will be comfortable with A solution that is within their budget. so for them simple luna < terra < sol < astra would suffice.
but if you want to use some LLM agent for anything commercial and you want to get the best and at the same time you don't want to overpay the easiest and the surest would be to create your own benchmark for your specific use case.
I recently experimented with some image generation models and I was surprised at how much context/prompting is needed to get something that doesn't look Lovecraftian. There was seemingly no downside to adding more and more detail up to the limit. Prose sits between code and images in terms of fuzziness and I wonder if promptmaxxing is the winning strategy for technical documentation.
I've been thinking that fine-tuning a model to write differently shouldn't be that hard (tedious, perhaps) and if I had the resources I would consider doing it myself.
> If you're not sure where to start, use GPT-6 Astra, our flagship model for complex reasoning and coding. Choose GPT-6.1 Sol to balance intelligence and cost, or GPT-6 Luna for cost-sensitive, high-volume workloads.
That page also seems to be updated after each new release.
Humans cannot be in the loop for autonomous work. I have agents review code and design/architecture documents, write alert messages based on rules, write analyses (e.g. outage incidents), and many other things. "Having a person write words" would mean not using agents, defeating the whole point.
I do not use agents to write on behalf of myself. I write my won technical write-ups and my own slide presentations. A lot of people don't; and it shows. There's much that agents can do better than humans now, but not everything.
> I'm all ears. Our intention to make things clear and simple, even if we miss the mark sometimes.
Thanks. I think the very response you wrote here is a better form of marketing for the software engineers than what openAI puts out there. When everything is groundbreaking and everything tops charts, nothing is.
I suspect a lot of target demographic (developers) prefer simpler and honest communication about not just the models high points but low points too. With the linked Terra post, if I saw 1 chart that clearly highlighted the usecases for which Terra is _worse_ than higher tier models, I would be more convinced overall.
Everything trying to occupy the top spot in my brain's limited "top model" representation space is not helpful.
While I generally agree that it'd be nice to just have "here is the Pareto optimal model for your use case. Here is the specifics we optimized each model for" so users don't have to do benchmarking and research as much, they're getting much better at explaining. It's basically what the commenter above said.
The website linked starts out with:
> We’re introducing GPT-6.1 Terra, a fast, efficient model that combines detailed visual reasoning and precise computer use alongside clearer, more natural writing. A significant step up from GPT‑5.6 Terra, it excels at the communication and computer-based tasks that fill the workday, from drafting proposals and interpreting technical drawings to filling out forms and editing videos.
If you go father down, it has graphs and details with specifics of this, including:
> We measured GPT-6.1 Terra’s computer-use capabilities... At medium reasoning effort, GPT-6.1 Terra scores 15 percentage points higher than GPT-5.6 Terra at about one-third the cost. It even performs on par with GPT-6 Astra at only one-tenth its price per token.
and
> On MMMU-Pro... It even nearly matches the scores of our most capable frontier model, GPT-6 Astra, and GPT-6.1 Sol at much lower cost.
The 4-tier naming is honestly as clear as you can get. I have no idea what people are so confused about.
Terra and Luna are dirt cheap and good enough for 50% (or more) of the typical junior dev's workload, and great for routine work. Sol and Astra have so much written about them at this point that it should be obvious what those are good at. The large/gigantic models for things that need full intelligence and slightly more branching thinking.
The obnoxious/confusing thing is that models are released individually rather than as a family.
Luna < Terra < Sol < Astra, fine. But does Terra 6.1 beat Sol 6? Does Fable 5.1 beat Opus 5.5? If I build something around this Terra 6.1 release, then when Luna 6.2 is released in a month, should I swap?
I understand the confusion, but isn't this kind of similar to asking does the iPhone 14 Max beat the iPhone 16 Mini? At least with phones there are hard specs we can use to compare, but in a lot of cases the answer will be "it depends"
At least there is a simple system - the words 'Mini' and 'Max' are intended to refer to the size of the item (that one is miniature, and the other is their maximum size).
If they were called iPhone 14 Luna and iPhone 16 Sol then this is less clear.
Or on the flipside, if the models were called GPT 6.1 Mini and GPT 6.1 Max, nobody would be arguing about the names being hard to grok. The words 'mini' or 'small' are more descriptive than 'Luna' or 'Terra'.
The problem with that is then you would have GPT 6.1 Max with max thinking which is different from GPT 6.1 Small with max thinking.
If you look back the naming problem started around the point that OpenAI started doing reasoning models (o1, 4o) where they had to describe this fork in model behavior that let a smaller model think longer and be smarter.
You can come up with naming schemes that don't clash, but there's something to be said for this approach where the thinking level is like you describe and the model name requires you to learn that Astra is the largest model.
1. Even more dead simple if it just used Moon instead of Luna, Sun instead of Sol, etc. - one less quick translation in the head.
2. Or even more dead simple if it just used "Small", "Large", etc. - removing another quick translation in the head. (On the other hand, all these suggestions are English - maybe Luna, etc. is more understandable for non-English speakers?)
You're right! The relative sizing of models is dead simple. This doesn't help when there are improvements between release numbers, and models are released one at a time instead of as a family.
When Sol 6.4 comes out and the latest Astra is still 6.1, which do you use?
I mean Astra actually means star, but putting that aside, we also have GPT sunburst and GPT flare - how does the logic apply to these?
Is a sunburst supposed to be bigger than a flare? I mean you get solar flares too? Or is a solar flare bigger than a sun 'burst' which usually just means some sun coming through clouds.
Sure, in a hypothetical world where the latest Max phone is two generations older than the latest Mini phone. In which case now you have to evaluate not only screen size but ALSO specs and component age and software support etc. How badly do you need the bigger screen? Etc
If you are comparing the iPhone 16 Max to the iPhone 16 Mini, everything is a lot easier.
Run your own evals then if you have this much of a need to "correctly" choose. That'll make it clear.
I'll just stick to mostly Sol on low/medium/high and start experimenting with Terra tomorrow onward to see what it does noticeably better than Luna/Deepseek flash.
Any pareto frontier model, especially from the leading labs is guarnateed to be good enough for whatever your expectations are at this point. A quick few day test should be enough for anyone to make their own conclusions.
Or I'm going to give up and not use any. Consumers aren't perfectly rational. This isn't like toothpaste, where, despite being overwhelmed at the store, I still need to brush my teeth so ahhhhhh fuckit and just pick a toothpaste out of the 30 available. The alternative is I keep using Google search and think AI is stupid. Or just use Facebook's thing.
The irony with tooth paste is that out of those 30, probably 28 are exactly identical. If all you want is fluoride tooth paste then pick the cheapest tooth paste.
If you want to go optimize your mouth hygiene it is actually more important to use dental bristles (flossing is not as good), because most of the work is done by the mechanical cleaning action of the tooth brush and you need to reach the places where the tooth brush cant get into.
Once you have done that, the next step is alcohol-free xylitol based mouth wash. Caries bacteria have trouble digesting xylitol so this reduces the presence of them in your mouth.
Only after you have done that, are you ready for something better than fluoride. Fluoride is a remineralization agent. Your saliva also contains calcium that helps your teeth remineralize. So the next thing you need is a more potent remineralization agent. Basically your tooth paste acts as a more concentrated saliva. Because of that, it is actually better to use e.g. fluoride in the morning and a different agent in the evening.
The classic options are bioglass (novamin) or hydroxyapatite. The latter is also available as mouth wash.
What would you want to see instead? Not sure about you but my needs are specific enough that I don't think OAI has evals for them, and I wouldn't want to give them that data even if they wanted to run it for me.
But here we have LLMs, wherein: It depends. The smaller models can in fact outperform the bigger ones, and get some stuff done cheaper, better, and faster.
If the question is "Which one is best?" then the answer is "It depends on what the fuck you're doing."
(If that answer is found to be unfulfilling, then let it serve as an indication that a person may be better-served by improving their questions.)
> Luna < Terra < Sol < Astra, fine. But does Terra 6.1 beat Sol 6? Does Fable 5.1 beat Opus 5.5? If I build something around this Terra 6.1 release, then when Luna 6.2 is released in a month, should I swap?
I understand how it can be confusing, but I don't think that this is exactly a problem.
Luna through Astra offer qualitative descriptions of the model's size. Presumably OpenAI trains its biggest model (Astra-ish, maybe even larger) first, then distills it to smaller sizes.
Verison numbers reflect something about the architecture or training dataset. Again, they keep these details close to their chest, but we should expect that weight-for-weight higher versions should beat lower versions.
This offers us two axes of 'better', but these axes are also not directly comparable. An earlier but bigger model may or may not beat a newer but smaller model. This is where 'spiky' intelligence comes in.
This is also where calls to 'run your own evaluation' really do start to matter. If you care enough about this performance[1], then you also have specific problems that you care about. No generic benchmark can fully answer that question, hence the need for a specific evaluation.
[1] — And you don't have to! A casual user is probably served reasonably well by anything put out in the past six months. I think Microsoft is still serving GPT-4o under some Copilot branding.
I think it is a problem, for OpenAI. If their answer to "which of your models are better" is to throw their hands in the air and say "fuck if we know, go figure it out yourself" then that's going to go poorly as soon as they try to actually sell to some of the market they're predicting they can sell to.
Whether they train the biggest model first or not is not relevant; what matters is what they release. Right now we've got 6.1 Sol and 6.0 Astra. Over at Anthropic they've got Opus 5.5 and Fable 5.1. The existence of a hypothetical Fable 5.5 that I'm not allowed to use doesn't help me.
- Understanding how effort and class interact (Is a high effort terra above a low effort Sol? Are larger models mostly always better regardless of effort?)
- understanding when (if ever) a new model tends to be better than an older but higher class one. Is an old astra better than a new terra?
If increasing effort and increasing model class produce improved results in a different way, I miss making it explicit in which way. In other words: What kind of issue warrants going up in effort? what kind of issue warrants going up in model class? What kind of issue is unlikely to be solved by either?
OpenAI itself cannot answer these questions. A prospective price-sensitive (or speed-sensitive) user has to experiment to find the cheapest (or fastest) model that is sufficient for each task.
Internal evals are still running but GPT-6.1 Terra is an AWESOME release for anyone using the API in agentic production applications where cost matters.
Personally, I pretty much exclusively use claude for coding work (sorry!). Our application is running almost 100% on OpenAI API (except for some fallback models). It is the best API on the market. The models are excellent particularly for any application that is cost sensitive.
Since internally you probably have a fairly good intuition for what these models do well, what about assembling a benchmark suite (with multiple dimensions) that produces results that show that intuition, but in a 'somewhat objective' way?
That could solve the 'more IQ' problem, because you'd actually have different dimensions (e.g. for 'raw intelligence', technical writing, emotional understanding, etc) that you could use to communicate the differences in models better.
(Of course people would instantly start complaining that the benchmarks numbers are probably fake etc, but it is what it is.)
I'm assuming you're not in the OAI marketing dept because leading with a simple thumbnail like "Luna < Terra < Sol < Astra" is helpful framing and dropping next-level bombs like:
> "Terra is probably better than Astra and Sol at technical communication, similar on vision and computer use (but much faster!), and a bit worse on most other things. For API developers, it's a nice midrange option if you want something cheaper than Sol but a lot smarter than Luna (especially for vision / computer use / communication). For subscribers, it's not a bombshell release - if you're hitting usage limits, maybe it helps you go a bit further, and if you want clearer communication or faster computer use, it can be a nice option to select."
Is just far too clear and useful! I'm kidding... but it's funny because it's a little bit true.
> Our intention is to make things clear and simple
Your post is the single best product marketing communication I've ever seen from OAI. Serious suggestion: maybe just say something like that on the website. Not buried in a blog post but above the fold.
Please can I ask about one parameter you didn't mention here? All of your models have both a reasoning effort setting and also a non-pro vs pro setting. I have some intuitive feel (from experimenting) for the difference a larger vs smaller model makes, and why that's different from adjusting reasoning levels, but I have no real feel for how pro vs non-pro is different in behaviour from adjusting effort.
I think you can think of Pro approximately as running multiple instances of a non-pro model, than giving all their outputs to a final judge model which combines them into a better one.
It's reasoning.mode, next to reasoning.effort in the responses API. Set to the string "pro" for pro mode. Sorry, I don't use first-party harnesses so have no idea what it maps to there.
> If you have suggestions on how to name our products better, I'm all ears. Our intention is to make things clear and simple, even if we miss the mark sometimes.
I'm not the one you were talking with but catching you is godsend since I just ran into a lot of confusion on the naming myself. I think some of it has to do with how they are labeled in the app.
I think simple numbers in naming is great, even 6.1 is fine bc I understand that if I want the latest model, I just need to look for the biggest number. Where I got confused is in the "work" tab, I saw 'Astra' is set as the default. I see the label for it is:
"Frontier intelligence for your most demanding tasks."
To me that reads "this is the 'best' model." Then I see Sol-6.1 which is labeled:
"For coding and every day work."
Then Terra-6.1 is:
"Terra 6.1 Codex Model" (tbh this one might be the most confusing, I'm aware of codex the product but Sol is the one labeled good at coding)
From my POV as a basic plan-based user wanting to do something like...
> design a robust event-driven system, here's the details [...]
I honestly have no idea which one I should use.
Would Astra-6 produce a better design than Sol-6.1? Is the '.1' a big step up from Sol-6? (I'm guessing no but I truly have no idea). I would assume the answer is Astra-6 would produce the best one, even though the labels don't imply it is necessarily geared towards coding?
But tbh the "default" seems to always work for me... and after reading your comment, typing this one out, being forced to think about it... it doesn't seem too bad given your challenges in communicating the purposes of each model and wide audience of consumers. I'm also sure anyone could find a million problems with decisions I've made in software I've built so I try to not be too harsh :)
It would be interesting to know which model is internally viewed as being best at jokes, I have a personal humor benchmark and so far models will either steal jokes (from comedians like Maria Bamford), or write something very unfunny.
When I was first learning to have agents interact I had a little script to orchestrate a model telling a joke, a critic collaborator helping develop material and an audience emoji rank them.
A favorite…
Prompt: “Tell a joke about mother in laws”
Gemma2 response: “What’s the best kind of mother in law? A dead one.” 0_o
Appreciate that these things have orthogonal measurements; with that in mind, given that the name signifies size, what does version number signify?
I.e. are all version numbers independent and Model X 7.5 is not necessarily of the same generation or cut off as Model Y 7.5? Or do all of your 6.1 models have something in common - cut off data, generation, baseline technology, approach, architecture, or something else?
These days, version number signifies a few things at once: the order of release, the relationship to other releases, and the magnitude of improvement. So something like 6.1 would indicate it was launched after 5.6, may share some similarities to other 6.1 models in terms of training data / personality (which is why it's not 5.7), and isn't a giant enough leap to merit the headlines and expectations that would come with a 7.
Back in the day we used numbers to also sometimes indicate size, which could make it extra confusing: e.g., we launched GPT-4.1 (an incremental update on GPT-4) after GPT-4.5 (a new model that was a big jump over GPT-4).
I actually do have a recommendation, which is to do something like a new mode (i know i know):
Auto xlow
Auto low
Auto med
Auto high
Auto xhigh
Auto max
With an optional flash mode.
Then as you release models, you pick the model and thinking level that you believe is pareto optimal for cost/performance.
So this auto updates each release, like Terra was probably off this list entirely, but maybe this would appear now in Auto medium or something like that.
I know from looking at charts that each model seems to have sweet spots (like I always use this model in this thinking level and if i want to go cheaper i go here). If you picked those for us, it would be interesting. I do think you should keep the model families for those wanting more consistent experiences, but for many devs they may want to just pick auto something and have it update over time. I think nice if it actually just said in parenthesis what it was, something like:
Auto Medium (Terra 6.1 High)
I am not sure if everyone would agree here though. I'm obivously advocating for more naming schemes, and i understand the irony.
It’s probably just unavoidable. Like when Apple release a new MacBook Air first and it is (in some ways) faster than the old MacBook Pro.
I think the planets thing is a bit confusing though. I hadn’t really thought about it being based on the size of the planet. I assumed it was based on how far away they were with terra therefore being the base one.
Thank You for OP asking the question and thank you parents for some insight.
>If you have suggestions on how to name our products better......
Too late for that. I hope you guys can at least stick to these naming scheme for at least another 2-3 years going forward. Not everyone follows AI / LLM intensely, and it seems there is a new naming changes every year or so.
At least conceptually OpenAI's naming is easier to understand than Claude.
It terms of capability, it would be great to make it clear somewhere that this is 6.1 Model "+" technical communication, vision and computer use improvements. So that at least we know what we should try with these new release and choose a better model for certain workload.
Right now, other than pricing, it is confusing as hell. And in hindsight I wonder if there are fatigue in choosing one people just go with Open Model and stick with it if it is good enough.
What makes this even more confusing is reasoning effort.
You have Luna, Terra, Sol and Astra, each with 4+ reasoning levels. That's ~20 options to choose from.
On a repetitive task like customer support or something, where you have decent benchmarks, you can plot cost versus performance, but you can't really do that on coding.
> In general, LLM performance is incredibly multidimensional. There's no guarantee of a fixed ordering where the model that is best at math is also the model that's best at jokes.
Maybe stop using version numbers then and just name models like you'd name children.
> Of course, model size doesn't capture all variation. Models that are trained later might have new data or new techniques or new architectures that earlier models lack. You can see this with models from all providers - Sonnet 5.5 outperforms Fable 5.1 in ways, and Gemini 3.8 Flash outperforms 3.1 Pro in ways.
Is this really so/could you give a rough example please? I've not used a lot of Claude or ChatGPT, but 3.1 Pro always feels most grounded and intelligent. 3.8 Flash (in my very limited testing) isn't bad, but it feels less nuanced.
(Fwiw I don't use it for code, but ask maths and real world logic based questions.)
Try not using putting two variables in the product name. Stagger the release and make sure you don’t release a more up to date version in the middle so that it looks like a “medium” model is more advanced than a “big” model. People see bigger numbers they think better.
Medium v10 vs Big v2 makes it look weird and confusing. Especially when there are a sea of influencers and twitters with vibe assessments who will tell you new models best olds models.
Can’t you just start at the top or bottom.
Release new big for paying customers -> release new medium for paying customers -> release new small for everyone
I like to use OpenAI models for ingesting a customer brief, usually a paragraph or two and then a read of a several customer files (eg spreadsheets and docs etc) and turning it into a system design blueprint. I'm currently using Astra for this phase but it's incredibly expensive relative to the other tasks that come after this. Would you recommend I stick to Astra or can I downgrade to say Sol or Terra for this job?
I'm just worried the system design won't be as good if I choose a lesser model so to speak.
Straight up - why not just name them Cheapest, Cheaper, Expensive and More Expensive? Cost seems to be the main driver of the intuitive difference between models.
Same reason that Apple doesn't name its series of phones the iPhone cheap, iPhone regular, iPhone semi-expensive, and iPhone expensive. It's poor branding, isn't easily extended to future models, and gets confusing when comparing across model generations, where an iPhone 14 expensive might be cheaper than an iPhone 16 cheap.
I love the current naming scheme and I am excited to see the astronomy trend expand in the future. I hope to see asteroids, binary stars, and solar systems at some point soon, as well as neutron stars, black holes, and eventually galaxies, and clusters of galaxies.
This is legally dangerous for the company because those words have generally well understood meanings, so a lawyer could argue that they promise a certain level of service. It’s safer to use vague words to describe the product variants.
> Models that are trained later might have new data or new techniques or new architectures that earlier models lack.
This seems like one of those cases, where CalVer would be appropriate, and you likely have ideas what you aim for while training, so technical model names (ids?) could include aliases to most things you said. Like
2026q4-xlarge-researcher, 2026q4-medium-humanitarian, 2026q4-medium-coder, 2027q1-small-analyst etc.
> I work at OpenAI and I can try to explain our logic [...] general, models that are bigger will be more performant.
> If you have suggestions on how to name our products better, I'm all ears. Our intention is to make things clear and simple, even if we miss the mark sometimes.
How about naming the models:
"Small", "Medium", "Large" and "Very Large"
Or refer to them by parameter count like other companies (e.g. this is 6.1 5T)
If I have a clothing store, and rather than name the clothes sizes Small, Medium and Large I decide to call them Urgon, Planfar and Fangor I shouldn't be surprised if people think my size naming is confusing or silly. And IMO it's not a great explanation to say "oh it's easy, if you are skinny then you should get Urgon, but then if you are a bit more of a medium build you should get a Planfar, but if you want something a bit baggier than Fangor will really work out for you"
The problem with small, medium, large naming is that they don't last. In a few years our names would have to advance to labels like giant, behemoth, gargantua. This is different than clothing, where we know that in five years humans will be most the same size as today.
As a point of comparison, GPT-3 was launched in a series of small, medium, large, and XL. Back then, XL meant a mere 1.3 billion parameters!
This kind of heuristic is actually extremely useful - and mostly what I try to develop over a couple weeks as the new models come out. I would love if we could get a fairly standardized take like this on the regular.
Of course we have far too many benchmarks (of questionable utility), but it would be helpful to get OpenAI's take on where it slots in, and how the landscape changes as these get rolled out.
This is kind of like Apple with the iPhone. You need 3 model products max. 3 is easily understood and people can be bothered (if they care) to track the difference. More than 3 is a mix of being redolent of ADHD (viz bad PR) while also being annoying for consumers to track the difference.
I can’t tell you the difference between the 10 different versions of the current iPhone and I can’t you the difference between the 5-6 different models put out by the Anthropics/OpenAIs of the world. At this point I just try to track the flagship and then the cheapest one or worst case I just use the flagship.
I’m sure internally there’s data showing people want choice and options but decision fatigue is of course very real.
Very fair. There's always a tradeoff in the number of products, and it's hard to make everyone happy. My suspicion is that the optimal number is probably a bit higher for developers on our API (who often build their own internal evals) than for individual users (who don't want to waste time re-optimizing every month). We're excited by Terra's uniquely fast computer use, but if it traction is poor, maybe we'll go back to 3. It all depends on the market signal - we only want to ship models people find uniquely useful.
"In the case of GPT-6.1 Terra, I'd say Terra is probably better than Astra and Sol at technical communication" -- you're doing the thing here too, underspecifying your own models: 6.1 Terra is better than…which Sol, 6.1, 6, 5.6, just the latter two, all of them? This 2D bluriness is a big part of the confusion.
I promise you this would become a giant mess. No one wants to pick GPT-good-at-writing-but-not-in-documents-and-decent-at-instruction-following-but-has-some-known-quirks-to-be-aware-of-p.s.-good-at-knock-knock-jokes-MAY-2027.
Even "writing" as a single concept is very multidimensional. There's fiction writing, nonfiction for experts, nonfiction for learners, deep research, quick lookup Q&A, copyediting, writing in documents / artifacts, instruction following about style, and default quirks (e.g., em dashes, negation, loving the word "tapestry"). Some models are great writers but less diverse, meaning people prefer the first story they read but disprefer the tenth. How does one label such a model?
> No one wants to pick GPT-good-at-writing-but-not-in-documents
But this is exactly what we have! Except it's worse, because the model usage recommendation isn't in the API, it's in some random person's X advice that's no longer current this week.
I understand AI companies wash their hands of this problem because it's hard, and they themselves barely understand the true capability matrix of the models they're selling.
I just don't think some effort to categorize models and expose that to users would be messier than what we have -- model selection driven by influencer voodoo bandwagons and analogies to literal astrology.
> important details would be obfuscated from end users.
I don't expect the vast majority of your users care. I would guess they want a rough proxy for "cost" and will assume cost=>capability are roughly linked in a linear fashion.
it is very possible that the users weighted by spend do, and I'm therefore simply wrong. But, 2c
Any news on when Terra gets into the hands of Free and Go users, maybe replacing 5.6 Terra?
Also, I think the naming is great. Like you said, Anthropic faces the same issues you do. Opus is much better at Fable in a lot of stuff without being quite as huge or as smart.
My guess is nothing, and they're currently just releasing them individually as soon they're ready instead of all at once so that they hit the market more quickly. AI moves quickly enough that a delay of a couple weeks could result in significant lost revenue.
Lower models typically use distilled outputs from higher model to improve .
all training runs have to at some level share the compute capacity ; while they have more than enough to run many large runs concurrently, they don’t have enough compute to run everything concurrently.
> As you note, our models are ordered: Luna < Terra < Sol < Astra
That is false, considering Astra and Luna don't even exist for 6.1. Terra didn't exist for 6.1 until yesterday. You don't get to make up imaginary tiers. The inconsistency wrt their availability, starting with day 1 of a new model version, is precisely the problem. OpenAI is a user-hostile joke of a firm that users would do well to abandon.
> If you have suggestions on how to name our products better, I'm all ears.
For some reason, this is just hilarious to me. Idk if you're being serious or if this is passive aggressive but the thought of OAI actually taking advice from a random HNer and basing their naming scheme off that is peak.
This is one of the last bastions of semi-serious discussion on the internet, and it's full of incredibly high level engineering talent.
Where else would you go? I'm honestly scratching my head thinking of someplace that wouldn't run you out on a rail, but that would give serious feedback. Kernel mailing lists are out, maybe there's something like that which I can't remember off the top of my head, but...
Also, nomenclature is incredibly hard. There are only two hard problems in computer science: cache expiration and naming things.
But regular people are not the consumers of OpenAI's API - they consume "Chatgpt" and don't care about model nomenclature.
The primary users of any API are the designers and engineers of the applications that use it. How many laypeople know what Postgresql is? How many of the applications they use run on it?
We aren't talking about the API. We're talking about the model names. If someone fires up the OpenAI website to ask it a question, nobody knows what the names mean. That's what normal people do and are probably super confused at what is going on.
I don't understand what's so confusing about this.
The lineup of models is still the same.
There are four models: luna, terra, sol, and astra.
Pricing increases for each level. This new 6.1 Terra is 50% cheaper than Sol 6.1.
They are lowering prices for newer releases, so yes if you compare Sol 6.1 to 5.6 Terra it won't look right, but why are you doing that? Look at the same release version for comparable numbers.
Each model improves in successive releases but they maintain pricing and capability coherency within a release version.
Yes, Sol 6.1 is supposed to rival Astra for some work, which is what you'd expect from a later versioned release model against a prior release bigger model.
Assuming they release Astra 6.1 at some point, I expect it'll be better than Sol 6.1, at a higher price.
I don't understand what people are asking for here? No reductions in pricing? Different model names on every version release? Not releasing any models for a version number until they're all ready?
Hey! I think it's confusing because there is a mix of three components:
- Model name
- Version number
- Effort level
All 3 of these are a one-dimensional balance of price/speed, and capability (Loose terms). The confusion is in how these overlap, how they're announced/marketed etc. So you go from 1 dimensions to 3; you can visualize it as a box. For a given task, how do you choose where in this box you select? Which regions are your favorite? Can you map each region of this 3D space to specific tasks? Label them with intuitive names?
Given the operations of the model are opaque, can you see how someone (most people?) might be confused?
But what's the ask? This is the cleanest way I see of them doing it.
I see a ton of criticism and not a single person coming up with anything better, unless it's to take away options, like removing effort level.
What about buying Intel CPUs? A car? Electrical components?
If you're not just using this at a consumer level to ask about cooking recipes (and if that's your use case, you really don't care about API pricing, or any of this, and can just use the default), and are using it professionally, then this seems like a low bar of effort to understand what you're buying.
I do grant you the relationship between model and effort level has an unclear relationship, but there are also tradeoffs in most complex technical products that are not always clear; this is exacerbated by the fact that this product is stochastic.
Why are engineers using this as a professional tool expecting this to be bought like a consumer gadget and then blaming the maker?
“The ask” is give me a clue how to choose the best settings for a specific task. Model releases happen so often that the calculation keeps changing. Maybe even OpenAI doesn’t always know the answer.
This morning I asked GPT which models and settings I should use to inspect my VM dashboard with computer control. It said:
- GPT-6 Astra at Medium
- GPT-6.1 Sol at Medium (alt)
- use Luna only when the route was already known.
But I had already completed the task twice with Luna and it was 5.6! It navigated through a few pages of my dashboard, pulled up the 30-day usage chart, analyzed my CPU and RAM, and explained why I didn’t need an upgrade. I’d give Luna 5.6 an A on this task.
Maybe OpenAI should give us a “Tiptronic” selector: recommend the model and reasoning level based on the prompt, while still letting users override it.
> “The ask” is give me a clue how to choose the best settings for a specific task.
What are you talking about? If it's for some slop you vibe check it. If it's for a serious effort you write an eval. Same as it has been for the last 3 years.
The OpenAI employee above says that some smaller models of the same generation are better at certain tasks than larger models, regardless of cost difference.
They should have a guide saying "Terra is best at technical documentation and communication, Astra is best at large software projects and research, Sol is good for development" or something like that.
Then he qualified to say it depends on the unique details of the documentation. I don't think its possible. You have to try all models with effort and compare all or go on reputation or vibes.
You are showing a selection from some third-party interface that mixes together models from six different suppliers: Mistral, Kimi, OpenAI, xAI, Anthropic and Google, with a bunch of old obsolete models.
Are you presenting this as evidence that OpenAI's naming scheme is bad?
If you don't want to spend the time to know the difference between all these different suppliers and their models, then why are you using whatever this program is that gives you so many choices, most of which are bad? Why is this OpenAI's fault?
There's also this "Instant" thingy which doesn't seem to be on the model-name axis but I really can't understand exactly what it is, even after asking ChatGPT about it.
Effort (as it's used in the context, how many "thinking" tokens to burn) is still a free variable. It's not always shown to the end user but you can set it independently for any of these models.
After that you just have model "strength" and version number. But of course strength is only comparable at equal version number and the two axes are almost parallel.
Anthropic also does this, perhaps to a lesser extent.
It's confusing because the models are released one at a time rather than as a family.
Let's say I have some task that Luna 6.1 cannot do, but Terra 6.1 can. So I build my product around that. In a month when Luna 6.2 is released, then I have to figure out whether it's worth to evaluate and swap or just wait for Terra 6.2.
Yes, but is that really “confusing”? The way they could solve your dilemma would be to wait to release Luna 6.2 until Terra 6.2 is ready, which just means you’d have fewer options and you’d have to wait. Which you’re free to do now!
Yes, doing releases this way means that it is confusing to compare model capabilities without going to enormous effort to run my own tests to determine something that should be knowable at a glance.
Waiting is the way to solve this. And despite the outcome being the same, making me feel like I'm being forced to choose to wait is a worse consumer experience.
I think a lot of confusion comes from the discombobulated rollout of these models as well.
They released gpt-6-sol (alongside astra and luna but no terra) and 7 days later released gpt-6.1-sol making gpt-6-sol basically DOA before any real adoption. Everyone knows it wasn’t just better RL/post-training, it was a completely different model underneath. Now they randomly reintroduce terra.
I’m sorry, but if you’re not even a tiny bit befuddled by this then you must be taking whatever OpenAI is having.
Part of the confusion is that it's not linear capabilities. One model is not strictly "better" than the other. They have orthogonal capabilities.
I think the names are more of a reference to price. Luna is always cheaper than terra which is cheaper than Sol which is cheaper than Astra (for the same model number).
But minor credit to Anthropic, at least they've mostly stuck to their four model names (Haiku, Sonnet, Opus, and Fable). You mostly know where a model is MEANT to be positioned from the name alone. OpenAI cannot keep it consistent from release to release.
But, yes, currently Opus Vs. Fable has the same issue as Sol-6.1 Vs. Astra.
I don’t think opus-5.5 is better than fable-5.1. I think opus is better overall because it’s much faster and cheaper than fable but if speed and cost is not a factor fable it’s probably just as if not slightly better than opus, especially at front-end design (tho the difference is marginal and no one should be really using fable day to day due to this alone).
> OpenAI cannot keep it consistent from release to release.
Hmm? They've been using the Luna/Terra/Sol/Astra scale for a few versions now and as far as I can tell it maps directly onto Haiku/Sonnet/Opus/Fable, in terms of the kind of thing it's meant to convey (if not in terms of the scale of differences).
Because of fear. Yoda was right, fear leads to anger, anger leads to hate. I see so much fear driving hatred for AI and the big labs on this site...it's really sad tbh.
It's like iPhone 17 being faster than iPhone 16 Pro. Because it's newer and benefits from advances in chip architecture, despite being a lower model tier. Sometimes different iPhone models are released at different times.
Both companies have screwed up their naming royally. Maybe that's intentional - I'm sure it doesn't hurt Anthropic to have people using Fable a bit longer than they need to and racking up a huge bill in the process.
They're always very clear about it when a new model comes out. You have to click away a banner to continue using the service. There's not much they can do to make it any more obvious without it being a major hassle for all the people who understand versioning.
It's like Intel i3/i5/i7/i9, or nvidia 50/60/70/80.
It is meant to give a rating inside a generation. Different generations are not necessarily comparable. an i7 is better than an i5, but not if you compare an i7 7xxx with an i5 14xxx.
A nvidia xx80 card is better than a xx60 card, but not if you compare a 1080 to a 5060.
The only difference for OpenAI is that not every generation has every model.
You're just not paying attention. If you run copilot in copilot using copilot effort, it performs much better than copilot in copilot using copilot effort. This is clearly explained if you visit copilot.com using copilot and then search for copilot.
If you daily all these models (and have a modest budget, not some >$400 corporate allowance), then it's pretty obvious that Luna is small and fast, and Sol 6.1 is the big one for jobs that need significantly above average intelligence, comparatively to a human engineer. I use both constantly. Astra is really not relevant for the average dev except to do some moonshot task with a vague verification loop.
Sure Sol 6.1 on low/medium can be considered "cheap" considering its abilities, but it's clearly not meant for the same type of tasks as Luna. You just grow an intuition for when something is solvable with a Luna-sized model. I'm guessing Terra is good for the tasks that you'd typically use Sol 6.1-low for, or for Luna type tasks (or other not-novel problem) where you want to reduce the risks of suboptimal decision making. It's exactly what I'd go with for feature additions where the correct solution is already mostly designed/specced-out.
The way to think of this is that (current) inferencing hardware has natural cut points of efficient scale (assuming some optimal architecture which at this point is almost always an MOE variant), so you always have an incentive to provide 3-4 model sizes (which nicely enough often means diff speed to answer also).
You are also rolling the training of your models by size and the lat est small model can sometimes beat an older model.
For a while we thought you just train the big model first then distill smaller models from it but since we've learnt the importance of on policy RL (basically the way a small model figures out a problem is very diff from a big model so it's best to RL optimize each model distinctly)
If you scale a model down, or switch out attention with linear attention, or introduce KV-compression, etc., you don't really know how it will affect model performance before you've trained the model.
And once you've spent millions training it, you might as well make it available if it's pareto optimal for some tasks. It also gives you more insight seeing how people use it in practice.
Agree, it's confusing. I hear people discussing which model to use for what tasks and my head starts to spin because there are so many properties to account for and it's generally hard to reason about LLMs.
"Speed, quality, price: pick two". I think I'd rather see a choice like that than the current convolution of $-per-token, sub quota, latency, simulated reasoning level, model size, context size, etc, etc.
>But then things go confusing, they released Sol-6.1 which sometimes is superior to Astra, and discontinued Terra which made it look like this
I don't want to rush to their defense here, but, when model families are getting refreshed every 6 weeks, there is going to be significant overlap, particularly with these staggered releases
Different classes and different generations. Isn't it roughly the same as what car manufacturers do? Or even Macbooks? 2014 macbook pro has worse performance than 2026 macbook air, this doesn't confuse consumers.
Wait I always thought terra was the smaller model and the further away you went in space the bigger the model (luna is further away so it’s better than terra)
It's especially confusing because Sol 6.1 is mostly just Astra but 10x slower at 1/10th the price - unless you pay for it to be as fast as Astra - at which point it's the same speed and price...
It's nice that they have Terra which seems to fill a void that's been needed since Sol 6 focused on cost (at the expense of quality), Sol 6.1 got a huge boost in quality, but became unusably slow unless you paid Astra prices - at which point, why not just use Astra?
the only confusing bit was dropping terra for a bit,
Luna (Small), Terra (Medium), Sol (Big), Astra (Bigger)
is perfectly clear, sol beats astra sometimes because Astra 6.1 was scrapped. and note that opus 5.5 beats fable 5.1 most of the time too without a skipped model.
None of the recent popular outages have been the result of a cyber attack or AI, and nobody credible is claiming otherwise.
It hasn't stopped people repeating "AI or Cyber Attack did it!" to one another in every thread, and getting more and more sure because others believe it baselessly.
reply