Top
Best
New

Posted by crorella 5 hours ago

GPT 6.1 Sol: Near-Astra intelligence for a fifth of the price(openai.com)
674 points | 607 commentspage 2
phpnode 5 hours ago|
What's driving the increase in release cadence here? We seem to get new models every week or so now, is this RSI?
az226 4 hours ago||
Mature training pipelines, plus ever expanding RL datasets of increased quality, and mega GPU clusters to finish training in a few weeks. Automated safety and reliability testing.
Aboutplants 5 hours ago|||
I do wonder if people switch back and forth between primary models (GPTvsClaude) that it may be a better idea to simply keep releasing updates as soon as possible in order to keep users from bouncing back and forth.
sockaddr 5 hours ago|||
This is it.

It's because they need subscription money and interaction data and so keeping a version bump in the wings to stop the bleeding from your competitor's version bump is the logical thing to do. It has nothing to do with RSI.

vividfrier 4 hours ago||
[dead]
pythonaut_16 4 hours ago||||
Maybe process maturity too.

Like think about a software org with good CI/CD versus one without. The mature org can do consistent incremental releases because each one is safe and low overhead, the messier org will do fewer big releases because each release requires a big effort on its own.

As model developers mature we might expect to see more frequent point releases rather than the big bang evolutions.

scrollop 5 hours ago||||
Probably one of the factors. Signed up to openai pro a few days ago, deciding between openai and anthropic, then sonnet 5.5 was released and am wondering whether I made a mistake.

Luckily it's not a mistake as now we have access to . . . dots.

(and sol 6.1, it seems)

geeky4qwerty 5 hours ago|||
jokes on me, I pay for all the subscriptions.
toasty228 5 hours ago|||
Opus 5.5 is better than they anticipated, it's faster, smarter, cheaper. I'm about to change provider for claude and I'm not the only one
copperx 4 hours ago|||
It feels like an updated 4.6. It's fantastic.
copperx 4 hours ago|||
> I'm not the only one

See, that's an/the issue. As soon as people start to flee to the improved model, they start to serve degraded models to keep up with the demand.

mckirk 5 hours ago|||
No, we're pacing ourselves to have the time to evaluate the impact each new model could have, obviously.
sharpshadow 5 hours ago|||
Response to DeepSeek’s technical paper and competition.
LPisGood 5 hours ago|||
Which paper are you referring to?
wg0 5 hours ago|||
What's that in summary?
Wheen 4 hours ago||
Not the person you're replying to, but judging by the emphasis on the cost of cached input tokens in the OP article, I'd guess it has to do with DeepSeek v4.1's KV cache efficiency. It uses <1000 bytes per token, so they're able to get 1M token context in under a GB.

Edit: https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash/blob/...

jonatron 5 hours ago|||
Probably just the singularity, no big deal
anotha_one 5 hours ago||
[dead]
MisterMunchkin 1 hour ago|||
Both labs are spying on each other and they get jelly when the other is releasing a new model, so they have to ship something at the same time so they don’t look bad.
orbital-decay 5 hours ago|||
Versions is marketing, snapshots/minor variations are easy and the number must go up. Release timing is another OAI's marketing tactic.

>RSI

Recursive improvement doesn't imply increased rate, another word for it is "iterative" but this probably sounds too boring to some people.

denysvitali 5 hours ago|||
They're pacing the frontier
blmarket 4 hours ago||
and seems like they're claiming Sol/Opus are not frontier (and only Astra/Fable are)
jchw 5 hours ago|||
It is the only way to reduce prices while making it look like a good thing.
lxgr 5 hours ago|||
Wanting to have the newer model than the competitor, presumably.
dandellion 5 hours ago||
The old "the bigger number is better", GPT announces model 6.1, the obvious thing to do next is to announce Gemini 27, and after that Claudé 3000, then a flute album.
lxgr 4 hours ago||
We swear, We Really Wanted To Make An "ASI" Model But This Is Literally The Way The Weights Dragged Us This Time
mynameisjonny_ 5 hours ago|||
The initial response to 6 Sol was bad, and Opus 5.5 was definitely winning the public vibes war. Makes sense to rush something out
motoboi 5 hours ago|||
New models are distill from the actual unrelease frontier models. They are just giving us better checkpoints.
jesse_dot_id 5 hours ago|||
No.
anotha_one 5 hours ago||
[dead]
SwabbyNat74 5 hours ago|||
Its a news cycle more than anything, and its ONLY going to get much, much worse. Daily releases, or multiple daily, 30-45, by EOY. Welcome to RSI!
agluszak 5 hours ago|||
They're releasing Sol 6.1 because 1. Astra 6.1 got postponed 2. Sol 6 is shitty 3. They have to release _something_ in response to Opus 5.5
tjwebbnorfolk 5 hours ago|||
Competition
mattnewton 5 hours ago|||
Anthropic’s IPO?
esafak 5 hours ago|||
Productivity is increasing as models get smarter; we are ascending the singularity. I'm serious.
colpabar 5 hours ago|||
What I don't understand is how much people have to say about every single one. Aren't we at the diminishing returns stage yet? Is there really that much to discuss?
infamouscow 5 hours ago||
If you look closely at various benchmarks, you'll see that often models will improve in certain areas while regressing in others. It suggests we're already at the point of diminishing returns.
system2 5 hours ago|||
Chinese model pressure. Many of my SWE friends switched to Chinese models. I also use QWEN and GLM for many of the api requiring projects and dropped OpenAI and Anthropic. The only reason was the cost.

EDIT: I love getting downvoted by openai and anthropic employees or their bots.

wg0 5 hours ago|||
I can't recommend Chinese models enough. My personal favorite is DeepSeek v4.1 Flash but I have tried Qwen 3.8, Kimi 3 and GLM 5.3 which are equally impressive but DeepSeek is the cheapest and fastest regularly hitting 270 token per second.

And yeah I have worked with Anthropic and OpenAI models, they're good but they cost a fortune while Chinese models are already really good at a fraction of the cost.

andybak 4 hours ago|||
DeepSeek v4.1 Flash is fascinating and uneven. It's way too chatty in OpenCode to be a collaboration partner. I tried dsh-tui which feels comparable to the codex/claude tui's and it's usable. but it seems to be "brilliant and yet stupid" in a way I can't quite put my finger on. I've got too much real work to get done to dig into it so until the big boys price me out of the market I'm back to my $100/month deal.
copperx 4 hours ago|||
I was working exclusively with DS 4.1 Flash until Opus 5.5 got me back to a sub. I was disillusioned with what was available.
thraway3837 1 hour ago|||
I keep hearing about these Chinese models, but what exactly are you doing with the models and coding? I have a need to fully write code with full tool calling capabilities. Not just methods or functions. I want to be able to prompt a feature and it makes the JIRA ticket, and fully implements it and makes a PR. I don't want to babysit it or even read the code. Once it creates the PR, I want it to monitor it for any comments fro Copilot/security review and then fix it as necessary.

Is that what the Chinese models are capable of? If so, how are you using them? API? Or is there an inference provider that is as fast as the big 2? What about the coding harness?

franzcoughka 5 hours ago||
[dead]
diego_sandoval 53 minutes ago||
I find GPT 6 to be lacking in common sense when it comes to interpreting my prompts.

I have to be more literal with it than with GPT 5.x, otherwise, it sometimes does something totally different than what I want.

Aboutplants 5 hours ago||
“OpenAI's new Pro 500 plan offers OpenAI's highest usage allowance and comes with access to its new "Ultrafast" feature — it also costs $500 per month.

At the same time, OpenAI is also making its existing $200 Pro plan less appealing. In Codex and Work, $200 Pro subscribers will see their included usage decrease from 20x of what the company offers to Plus users, down to 10x of that same allowance. In ChatGPT, meanwhile, GPT-6 Pro message caps will decrease from 200 to 100 per week.”

https://www.engadget.com/2272106/openai-adds-dollar500-pro-s...

Yikes

TomGarden 5 hours ago||
They're really (finally?) starting to behave like a company bleeding money.

Our VC-backed subscription days are numbered

glaslong 4 hours ago|||
Alas, I did enjoy burning investor money on my taxis, movies and tokens.
onlyrealcuzzo 3 hours ago||||
> Our VC-backed subscription days are numbered

Well, the time it takes to compress frontier intelligence down to DeepSeek V4.1 Flash costs (basically too cheap to meter) is dropping, and the differential between the two is also dropping...

So... who cares?

m3kw9 4 hours ago||||
I'm ok with whatever price they give out given they are not a monopoly and have competition, the lock in is minimum for me. This means they have legit reasons to send us this price plan. I don't believe they would shoot themselves in the foot when there is cut throat competition (Claude/opensource) out there.

Lastly, I'd like to actually use it in the real world to see how far my plan goes or if its unusable.

LeBit 4 hours ago|||
Let’s pray Chinese models are not banned.
Madmallard 3 hours ago||
how could u even ban them? lol
girvo 2 hours ago||
Same way they’ve banned a lot of Chinese networking hardware: make it impossible for companies to use it.
killingtime74 1 hour ago||
anyone can run it on their own laptop
user43928 4 hours ago|||
Unfortunate that the Ultrafast is only available with the $500 subscription.

Tibo said that the existing $200 subscriptions keep the 20x factor for a while.

Ultrafast would have been nice with the temporary "Pro 400" plan.

cactusplant7374 3 hours ago||
Ultrafast uses 6x the usage. They probably realize that people will complain if the plan limits are too low. In any case, the TCO of the newer chips is supposedly lower. Hopefully everyone is on ultrafast eventually.
torginus 4 hours ago|||
I think that's by design - they're going to IPO soon so if they can get a significant percentage of users to switch from the $200 to the $500, they can 2.5x projected revenue.
glub 4 hours ago|||
Yeah, that's not going to happen. They are more likely to lose a lot of customers, unless Anthropic does the same thing.

But $200 is likely the ceiling of what people will pay for a subscription with usage based on vibes.

latentsea 4 hours ago|||
For consumers they may as well buy GPUs and run local models. The cost is same over a year or two but infinite token usage, they get to keep the hardware, and local models continue to improve over that time too. I can't justify $200 on SOTA models for a personal subscription after Qwen3.8-27B. And it's only getting better from here.
glub 4 hours ago||
Yes, either US AI corps reduce the cost of their top tier personal subscriptions down to what people are already paying for other expensive personal apps (e.g. Adobe), so ~$50-100, or open weights are going to eat their lunch very quickly. We're not there yet, as current hardware doesn't allow you to do things like multiple parallel agents, but we'll get there soon enough.

$500 for the old $200 is definitely a fumble.

latentsea 4 hours ago||
I have multiple GPUs now as a way to solve that.
seizethecheese 2 hours ago|||
People said the same about $200 a month. I think the ceiling is probably much higher. Companies regularly spend 10% or more of employee cost on offices, SaaS, equipment. I could see these costs going to 10% of white collar income.
glub 2 hours ago||
> People said the same about $200 a month

This is missing an important context. And I actually remember this well, because I was saying that too. And the reason I was saying is that $200 plan didn't come with API usage, it was a chat plan.

It made no sense up until they started including API usage. Just as $500 makes no sense now.

> costs going to 10% of white collar income.

There's a permanent and ever lowering ceiling maintained by open weight models. It makes no sense to justify paying 10% of income permanently for something that will get you unlimited local inference for a 6 month subscription cost.

seizethecheese 2 hours ago||
I’m not quite sure I understand the API usage point as it relates to regular customers.
glub 2 hours ago||
You could only use it on chatgpt.com

Now you can use it in coding harnesses that call the API.

adonese 4 hours ago||||
Very risky to do so especially considering how well is opus 5.5.
scottLobster 4 hours ago||
You think these guys care about risk?
Computer0 1 hour ago|||
I am skeptical that individuals on the $200 and $500 plans make up that meaningful of a portion of revenue.
moregrist 5 hours ago|||
This is pretty typical product positioning. You want to sell to both high-end and low-end users, so you offer products at a few price points. Then it turns out that that middle is a much better fit for most users. So you start making the middle a worse fit to push most of those users into the higher tiers.

Long term, this only works if you have a non-commodity, and if the higher tier is actually more profitable. We'll eventually learn whether both are true. For OpenAI right now, it's probably enough to just increase revenue, even if the higher tier is even less profitable.

5555watch 4 hours ago||
The 200$ plan was appealing because you got 4x usage for 2x the price.

Now, as it's linear, it makes much more sense to downgrade to 100$ OAI and pick up a 100$ Claude sub. (without doing the numbers) the usage should remain the same, total paid the same, but having access to best of both worlds. It should be a win for the user, and a loss for OAI.

With this in mind, it sounds like a fumble by OAI.

jpadkins 3 hours ago||
This is what I did. Hope it works out. The other benefit is you have a more natural method to avoid lock in. A lot of "improvements" to the agent harness I believe are attempts to build customer lock in.
latentsea 4 hours ago|||
At $500 per month, it's cheaper to just buy GPUs and use local models.
tripleee 4 hours ago||
Have you looked at the prices of GPUs lately?
latentsea 4 hours ago||
Yup. I got an R9700 recently for exactly this reason. Figured if I'm going to spend $2400 a year I may as well have something to show for it at the end of it.

That they are expensive and climbing doesn't negate my point if the cost of the subscription over how long you plan to keep it is equally or more expensive than the GPUs. You can put together dual 5060 Ti or 5070 Ti systems to run local LLMs too. You don't need to splurge on a 5090. That's a bad option at this point.

tripleee 3 hours ago||
What models are you running locally? Are you banking on them improving or do you think they're good enough today? 32GB of VRAM there wouldn't be close to enough to run the best local models.

I've messed around with Qwen3.6-27B but I'm not sure if it could yet even replace Luna for me.

ndbe 4 hours ago|||
[dead]
honkycat 5 hours ago|||
Wow, canceling my sub. Lets see how Claude is doing these days.

I can justify $200/mo but more than double is not appealing to me.

WinstonSmith84 4 hours ago|||
Well, here is a breaking-news for you: the 20x from Claude is not a 20x on the weekly usage, it's a 20x on the 5h usage, while the weekly usage is simply double the $100 plan...

Basically OpenAI aligned with Anthropic on the weekly usage with the caveat that OpenAI doesn't have a 5h limit.

MCArth 4 hours ago|||
If you've used both you know the OpenAI plans don't compare to Anthropic plans _at all_. Claude code subscriptions are probably worth 4x as much in API spend compared to the same OpenAI subscription tier.
nostrebored 3 hours ago|||
I think you have probably started using OpenAI recently -- one draw used to be that it was really, really hard to ever hit limits. If you did, you probably had usage resets available.

I think this is still true provided you're not using Astra.

machomaster 1 hour ago||
The shitty thing about OpenAI's resets is that, unlike Anthropic, they also reset the limit (on the next natural weekly reset). It means that of you pushed the reset button 5 days into the week, you only get 2 days' (2/7 of weekly) worth of extra tokens.
andriy_koval 2 hours ago|||
people say this, but I am wondering if there is benchmark/dashboard which actually measures this?
the_duke 3 hours ago||||
It used to be bad, but right now with the 200$ Claude sub I find it pretty hard to blow past the session limit.

You have to do a lot of things in parallel.

InsideOutSanta 3 hours ago||
Yeah, Fable is essentially unusable, it just burns through quota, but Opus 5.5 is great. The $200 plan goes a long way.
diffuse_l 4 hours ago||||
OpenAI 20x wasn't 20x even before that change. I got a lot more from Claude 5x than Codex 20x...
spiderice 4 hours ago||
You are literally completely flipping reality. Codex was, in fact, 20x. It was Claude that was not 20x until they got caught.
diffuse_l 4 hours ago||
I'm describing what I got from 20x Codex vs Claude 5x. Codex is just not worth the money, at least for me. What's flipped is the value you get for each of those
enraged_camel 4 hours ago||||
You are painting half of the picture, perhaps on purpose? The other half is this: Opus 5.5 is significantly better than both Sol 6.1 and Astra, and with the newly increased limits across the board, it is quite difficult to run out (unless you're spamming agents at Max effort). So it is a much, much better deal than OpenAI's Pro 100.
WinstonSmith84 4 hours ago|||
> Opus 5.5 is significantly better than (..) Sol 6.1

Come on .. this is barely released and you can already make that assessment?

And no, the $200 Anthropic plan is not significantly better than the $200 OpenAI plan, it's just the same Marketing non-sense and anybody shall now rather stick to the $100 plan of both of these provider if the monthly budget is $200. Anthropic doesn't have a Luna Max equivalent, and frankly Sol 6.1 is yet to be thoroughly tested.

cmrdporcupine 4 hours ago|||
"For antitrust reasons, it’s helpful for the US government to mediate or at least enable these discussions — they don’t need to participate, but do need to issue a narrow waiver for certain kinds of safety conversations. " - Dario a couple weeks ago.

Yes, he was talking about safety, but IMHO they're likely already IMHO pushing the boundaries of cartel type behaviour. And they will use safety as the cover to make it happen.

I suspect we'll see serious price fixing and the DOJ do nothing about it because of the inroads these people have with the Trump regime.

Whether that survives contact with Chinese open weight models is hard to say.

spiderice 4 hours ago|||
> According to the company, existing subscribers will keep their current limits for a time, and will later receive a one-time credit to help them make the most of their new reduced allowances

Might want to hold off on canceling and continue to bleed them dry until the nerf hits

honkycat 1 hour ago||
I just got an email telling me this isn't true. They're immediately cutting my 200, which I've had for like a year.
surgical_fire 5 hours ago||
OpenAI is deeply unprofitable, particularly on those pro plans.

The only way is for prices to go up. Way up.

mrtesthah 4 hours ago||
It really does look like OpenAI is trying to gradually get rid of their subscription plans. Every week there is noticeably less usage available to them while each new model release boasts substantially cheaper API token pricing. If this continues then the two pricing models will eventually be at parity.
surgical_fire 2 hours ago||
The subscription plans are a huge money sink, that obviously will have to go away or be priced at a ridiculous level to make sense.
machomaster 1 hour ago||
This is not true at all, at the most fundamental level. There is a reason why all the businesses (IT, gyms, cars, restaurants, streaming services, music, games, stores, apps, food delivery, magazines, newspapers, shaving blades, parfume, etc) are doing everything in their to get subscribers and are willing to decrease prices in order to get customers who are paying the monthly (or even better, a yearly) fee.
surgical_fire 13 minutes ago||
This is delusional.

The cost of providing the tokens for a heavy user (and let's be frank, the people paying $200 are likely heavy users) is many, many times more than the $200 recurring revenue they generate.

A_D_E_P_T 5 hours ago||
Looking at the token prices, if this is half as good as 6-Astra for 3D model creation in Blender, it's going to be an absolute game changer.

Opus 5.5 is definitely better at coding, but nothing even comes close to 6-Astra for work in 3D graphics...

CuriouslyC 5 hours ago||
From the results of a lot of YouTubers in the space, I think Opus 5.5 is pretty competitive with Astra in 3D. It's slightly worse at spatial detail but better at aesthetics and little touches.
A_D_E_P_T 4 hours ago||
Interesting! Can you share an example?
CuriouslyC 3 hours ago||
https://www.youtube.com/@stefan_3d_ai

A number of others have done game/3d video benchmarks but this guy is probably the most prolific.

kroaton 2 hours ago||
That dude is a grifter. His German friend is even worse.
ekun 5 hours ago|||
How is it with animations?

I have played around a little bit with fixing some rigging problems and was impressed, but Opus even warned me it was bad at animations cause it can only really grab screenshots to process static content.

godwinson__4-8 5 hours ago|||
You need to use the Blender MCP. There is an official plugin for this now, so the third party one can be avoided.

I've only dabbled but yes with SOTA models it is very good at animating and really most Blender tasks you can think of. Certainly if you are coming at Blender at below expert level it makes it far more accessible and fun to work with.

There are still rough edges of course. But try the official MCP out with Astra and judge for yourself.

A_D_E_P_T 4 hours ago|||
I've only tried animating models in Astra-6, and I was quite impressed! It's rarely able to one-shot things perfectly, but it usually gets pretty close.
lukan 5 hours ago|||
Have you tried fable? (I did small experiements and was satisfied, but maybe there are reasons to switch?)
A_D_E_P_T 4 hours ago||
No, because I always hit my Fable quota (Max 20x) in 12 hours on simpler tasks, and I'd hate to need to buy tokens at API pricing.
therealdrag0 4 hours ago||
After all the hype, I’ve been kinda disappointed tbh. Modeling specific models are so much better (eg. Tripo3d). Astra still models some janky crap for me.
jdprgm 3 hours ago||
6.0 Sol was literally a week ago... Basically continuous integration for model releases at this point.

Since Luna is so dirt cheap compared to Sol/Astra it would be nice if they could set or you could reserve some small percent like 3-5% of usage pool on codex just for Luna so if you hit usage limits you can at least still run a lot of Luna.

alright2565 1 hour ago||
They do, when I was on the $20 plan, I got shown a Luna Reserve model which had its own dedicated quota.
bayesianbot 2 hours ago|||
Interesting idea, but at the same time it is just so cheap that you can just run it with API pricing. I sometimes do even if I have available usage that I'm going to cap so I'll save it for bigger models
poisonborz 2 hours ago||
As many have surmised, this may have been a panic rename of Astra 6.1
machomaster 1 hour ago||
Astra 6.1 light, not the full-fledged version.
aabajian 3 hours ago||
Opus 5.5 is on another level, especially when it comes to mathematics implementations. You can drop it a PhD-level physical simulation (for example, a contrast-injection simulation for angiography in my case), and it just...implements it. With full-on WebGL rendering in the browser, from scratch (or using an existing library, if you prefer).
fraywing 5 hours ago||
> GPT‑6.1 Sol matches GPT‑6 Astra at roughly one-fifth of the cost

Astra is a pretty impressive model. Excited to try this.

gobdovan 4 hours ago|
They have also cut allowances for subscriptions in half. So even in the best case scenario it's about 2.5 times cheaper for Codex users. They just seem to have matched Claude Sonnet 5.5 *API pricing*, but from what I see online, it seems Claude Code now has a much more generous subscription allowance.
Tadpole9181 3 hours ago||
Only for the $100 subscription, correct?
gobdovan 2 hours ago||
Only for the $200 one. The $100 one was already pretty poor value for allowance/$. Without the old $200 sub, I wouldn't have used Codex.
jumploops 2 hours ago||
If the Terminal Bench 4.0 scores are to be believed[0] GPT-6.1 is an incredibly efficient model.

Yes, benchmarks aren't real work blah blah, but the delta here is so large compared to Astra, it makes it seem like this is distilled Bel or similar.

[0]https://x.com/thsottiaux/status/2105007628460109953

modeless 5 hours ago||
GPT 6 Sol is obsolete after only one week! I am glad that they are not afraid to update the models more frequently. The Navier-Stokes thing revealed that it took them only a week or two to train a model more capable than Astra, and I want the pace of public releases to keep up with that.
pmdr 4 hours ago|
I had it write some code the other day, boy was it awful-looking compared to 5.6. Worked perfectly, but ugly nonetheless.
samuelknight 4 hours ago||
Sol 6 was a flop. Nobody would have cared if it was called Terra 6.
TomGarden 5 hours ago|
Impressive improvements, but GPT 6 Sol came out 7 days ago, and this one will behave differently. The panicked pace is becoming a liability, maybe they should have waited and released this as the 6.0 release
enraged_camel 4 hours ago|
They are behind, hence the panic. On top of that, Sam has been trying to do another funding round, so he's desperate to make the company look good.

Opus 5.5 was a gut punch and my impression is OpenAI is still reeling.

atonse 4 hours ago||
People said Astra was a gut punch and that Anthropic was reeling. (Opus 5 was almost universally panned)

The best thing is that we benefit from these constant back and forth gut punches :)

jhonof 3 hours ago|||
I actually heard the opposite, the folks at Anthropic felt pretty confident they were ahead after the Astra release because it wasn't as good as they were expecting.
JacobAsmuth 3 hours ago|||
I don't know why anyone was saying that when Anthropic clearly knew Opus 5.5 significantly outperformed Astra at the time of Astra's launch. I think it might be a good exercise to go back and find out who called Astra a "gut punch" and lower your credence in their future claims.
More comments...