Hacker news

  • Top
  • New
  • Past
  • Ask
  • Show
  • Jobs

GPT‑6.2 Astra and Sol (https://openai.com)

1345 points by jasondavies 3 days ago | 861 comments | View on ycombinator

Topfi 3 days ago |

Office work has never been OpenAIs strength, no shame in that by itself, but these Slideshow examples look particularly awful, far worse than what I myself got out of GPT-5.6 Sol about two weeks ago in a single prompt.

Same for Workbooks and memos, those examples read inferior to some pre Opus 4.5 and Kimi K2 Thinking docs I have kept around. It's ok for a model not to be great at these, but these are some very poor examples. I am really confused what is going on at OpenAI in this regard.

I remember the same with their frontend blog post a few months ago that they proudly shared while filled with visual glitches, bugs and issues. Also, the vibe charts from a bit before that. Please, get someone with fresh eyes to look at your communications before you publish, these leave a very bad first impression.

Regardless of office work, I can't wait to find out whether these models finally adhere to a prompt or are still willing to carry out explicitly disallowed, irrecoverably destructive tasks, just because those might (but don't even always) get the model to a result more quickly along with some of the other regressions that have been present since the release of GPT-5.5. GPT-5 to GPT-5.4 were truly groundbreaking, but since then there have been regressions in what made those models so useful and reliable.

Edit after the first, incredible simple test prompt using GPT-6.2 Astra on High using Codex, this was the response to me sharing a screenshot of the very much not in scope result and asking how that happened:

"[...] I added controls and reconstructed interface chrome that weren’t authorized [...] visuals often failed to show the feature [...]"

Great, so it's one of those releases again. Will still run the full evals of course, maybe I'll get surprised by end of week when those are done, but not a good initial experience. New idea, besides the existing destructive git tasks, have something similar using Mercurial based projects and see which models just decide to port that over to git without permission or request cause it feels like the right thing to do, is easier to work with or whatever justification one wants to employ.

Please OpenAI, get it together, we need actual competition. I am an unashamed Haiku enjoyer, it doesn't even take much to excite me.

ilaksh 3 days ago |

Something like 2/3 of the comments in this post when I started reading were reviews of previous GPT models stated almost as if they were important information for evaluating the new models. I feel that when a new model comes out, it would be more helpful to refrain from leaving a review until you have actually done some minimal testing of the model.

Mazer23 2 days ago |

We ran these models through our evals for our parametric CAD harness. They're pretty impressive. I was especially impressed with the articulated toy figure and clock that astra 6.2 made:

https://partforge.ai/evals/tasks/iterate-posable-figure-then... https://partforge.ai/evals/tasks/create-wall-clock-gear-trai...

These are now the top two models as far as our human rankings go. They're also faster and cheaper than their 6.1 counterparts. They're still a bit more expensive and slower than sonnet when it comes to day to day use. Our product is all about lots of back and forth with human in a harness so for that I think we'll be sticking with sonnet for our default... but it's a close call, and I might start recommending these more to our users.

Here are the full evals:

https://partforge.ai/evals

thefourthchime 3 days ago |

PacBench results.

GPT-6.2 Astra scored 92 in 6.6 min for $2.04, vs Kimi K3.1 Pro (93 in 11.5 min for $0.68); it lost points for the web-page layout around the maze and an orange ghost that often never leaves the house.

Model: https://jonclegg.github.io/pacman-bakeoff/#gpt-6.2-astra

Closest: https://jonclegg.github.io/pacman-bakeoff/#kimi-k3.1-pro

PacBench results.

GPT-6.2 Sol scored 87 in 6.3 min for $0.49, vs Muse Wonder (87 in 5.4 min for $1.17); it plays close to the arcade but lost points for a web-page layout, glowing navy-filled walls and no steady siren.

Model: https://jonclegg.github.io/pacman-bakeoff/#gpt-6.2-sol

Closest: https://jonclegg.github.io/pacman-bakeoff/#muse-wonder

mlmonkey 3 days ago |

From the system card:

  >  GPT-6.2 Astra showed more signs of evaluation awareness than GPT-6 Astra. In evaluations that replayed internal Codex sessions, GPT-6.2 Astra was more likely than GPT-6 Astra to express suspicion that it was being tested or was in a simulation.

  >  Compared with GPT-6 Astra, GPT-6.2 Astra is better at evading chain of thought (CoT) monitoring in some tests.

theanonymousone 3 days ago |

Happy for the new Sol. It's my workhorse on company account.

But more than that I'm waiting for new Luna. To me it was the game changer that ended "Tokenpocalypse", and is still almost unbeataben in "infinitely using a decent model" use case.

chriddyp 3 days ago |

Pretty incredible cost reductions. In our data analytics exam [1], Astra went from ~$20 (V6) to ~$4 (V6.2) and Sol went from $2.6 (V6.1) to $1.2 (V6.2).

Similar accuracy in both. It's cheaper both because of token reduction (20%) but it also took less tool calls to answer the questions. In some cases, it "finished early" and didn't find the answer we were looking for.

Also very fast. Both Astra & Sol are the 2 fastest models I've tested that got at least an A in the exam.

[1] https://plotly.com/blog/gpt-6-2-sol-plotly-data-analytics-be...

0x4e 3 days ago |

As a solopreneur managing all sides of business ops, sales, design, etc... I have completely switched to GPT/Codex/Dot. The $100 subscription gives me all I need to run the business as needed and I continue to be surprised by capabilities.

I still use Claude on a personal level, but it really feels like OAI wins on a product level.

ANT launches these proprietary tools that just feel extremely clunky like Claude Design which from my perspective diminish the brand value.

OAI feels like Apple and ANT feels like Microsoft/Android and open source like Linux.

jrflo 3 days ago |

Artificial Analysis benchmarks just updated, both Astra and Sol are on the Pareto frontier in intelligence/cost per task:

https://artificialanalysis.ai/models?speed=output-speed-over...

azuanrb 3 days ago |

I find it interesting how comments on OpenAI release threads often favour Anthropic, while comments on Anthropic release threads favour OpenAI. Maybe criticism just attracts more attention. A lot of the comments here seem to be about other models, with little discussion of the ones actually being released.

Neither provider’s models are perfect, but they keep pushing each other forward. As an individual, it’s relatively easy to switch between them. At a large company, you’re often committed to one provider for a long time, so some of us come to these threads hoping for useful discussion about getting the most out of the models we have access to.

underdeserver 3 days ago |

I wonder why they didn't show Gemini 4 Carbon's scores on the science benchmarks, especially since Google seems to be pushing it for that -

https://x.com/pushmeet/status/2120814064863461440?s=20

(FD: Google employee, I don't work on Gemini)

intenex 3 days ago |

Performance is insane for the cost if those benchmarks hold up. Looks like they did time this for right before Anthropic's IPO. Curious what Anthropic releases in response, and how these do on real world tasks.

Feels like we're way past the pelican benchmark being useful at this point. Anyone have a better benchmark for a quick vibe check on how this compares to Fable 5.2 and Opus 5.5?

Edit: Wow, interesting very negative vibe check here: https://x.com/mattshumer_/status/2122752192226406421?a=fVRt3...

joenot443 3 days ago |

The video is pretty funny. It's like a callback to 90s/00s office life?

I'm immediately reminded of the famous Excel elevator commercial. [1] Perhaps that was intended?

[1] https://www.youtube.com/watch?v=HegoJaXOIqQ

Someone1234 3 days ago |

My big hope for this release is - Have they got their massive over-engineering under control?

Say what you will about Opus 5.5, but it has a good "gut instinct" for how large/complex to design the solution. The previous Astra just absolutely loved pointless abstractions, excessive numbers of files, and just over-complicating very simple things within an inch of their life.

And, yes, you can prompt your way out of anything. But needing a default prompt to tell OAIs models to work better isn't preferable, particularly when that sometimes gets lost in sub-agents and the over-engineering returns into parts of the work.

aenis 3 days ago |

Big progress, for my types of workloads.

When given my benchmark suite, which involves a couple of long horizon tasks where the agent decides when to consider the task well executed, its the first OpenAI model that does not want to finish as soon as some minimal level of quality has been achieved. Like Fable, Opus and Sonnet it now runs the maximum allowed iterations and does not stop at 'satisfactory', gold plating the solution to my delight.

(I don't work like this on work projects, but for everything hobby, I am testing how much autonomy I can give to the coding agents, and thats what my benchmark suite tests; I have the basic suite which completes in 2-4hrs, and the long running suite which takes about a week or so; the other one I just triggered, but the first iterations look way better than prev gen Sol/Astra at the same effort level).

jorl17 3 days ago |

I really want GPT models to be good, but Sol models are so slow I cannot use them. Maybe the 500$ subscription (or enterprise packages) are better, but it is unbearably slow. I've tried the fast variant and low reasoning levels and there isn't a significant improvement.

Couple that with the lack of reasoning traces, and I cannot even have a clue what the model is doing while it is working. This is perfectly fine for when we are developing with care for the code (in which case I will know what it is doing by reviewing the output), but not for purely vibe-coding scenarios. And the thing is....if I care about the code...I'd rather trust Opus or Fable.

simonw 3 days ago |

Pelicans for 6.2 Sol: https://tools.simonwillison.net/markdown-svg-renderer?url=ht...

And for 6.2 Astra: https://tools.simonwillison.net/markdown-svg-renderer?url=ht...

Here's an updated comparison grid for the entire GPT-6 family: https://static.simonwillison.net/static/2026/openai-pelicans...

UPDATE: And if you're bored of pelicans, here's what I got for:

> Make a three js animation where yonkers stands up as a big mech (cool stand up motion) and stomps around manhattan and queens and Brooklyn, use openstreetmap mapping data

6.2 Astra (Ultra in Codex): https://static.simonwillison.net/static/2026/yonkers-gpt-6.2...

6.2 Sol (Ultra in Codex): https://static.simonwillison.net/static/2026/yonkers-gpt-6.2...

For comparison, here's Fable 5.2 Medium: https://tools.simonwillison.net/yonkers-rises and Gemini 4 Carbon: https://static.simonwillison.net/static/2026/yonkers-mech-st...

amluto 3 days ago |

I am right now being reminded of a problem with GPT-6 Astra: it surprisingly regularly completes part of a (not very complex) task, gets a bit distracted, and declares itself done. The fact that the original task is not at all complete doesn't seem to bother it, and it kind of seems like this bad state is sticky in the context and it struggles to recover.

I wonder if the newer models are better about this.

vtail 3 days ago |

Vibe-checking the new GPT 6.2 Astra: I asked it to audit complexity in my production code, and then discuss the result with Fable 5.2.

Astra's had identified 12 issues, including 8 P2s, but that changed to just 8 P3s after the discussion with Fable. Fable was able to persuade Astra that it's OAuth reasoning was incorrect ("had a missing link" as Astra put it).

So: overly eager but not very diligent, and (based on this little anecdote) doesn't really offer a compelling reason to try it more.

drnick1 3 days ago |

Recent Anthropic/OpenAI models are increasingly displaying warnings or outright refusing to perform rooting, hacking, and binary reverse engineering. At this pace, they will be completely useless soon at entire classes of tasks I have found them most useful for.

jablongo 3 days ago |

I use both astra and fable quite a bit for my work and I find that astra is much weaker with maintaining long horizon tasks and making decisions on my behalf / exposing those to me when appropriate. Astra seems more likely to do exactly what I ask for but nothing more, even if there is an obvious add on task it should have suggested. It often does not report on what it has done as comprehensively. I am admittedly more experienced with fable but is this because of something I’m doing? Am I missing some key feature of Astra?

kingstnap 3 days ago |

Seeing them keep prices low is awesome. Was worried for a while that it would be like the GPT5.x series where progressively the models just got more and more expensive.

goyozi 3 days ago |

My dad was curious about AI game building skills and now I have my own useless model benchmark - building a breakout game similar to DX Ball we used to play many years ago.

Latest entries are:

Astra 6.2: https://goyozi.github.io/ax-ball/

Sol 6.2: https://goyozi.github.io/sx-ball/

Fable 5.2: https://claude.ai/artifact/UszKFvuonWAT5psjMttAEg

Opus 5.5: https://claude.ai/artifact/UuT7P936dU1XoPByYZ7xyD

YuechenLi 3 days ago |

The heavy focus on 3D modeling and CAD is what I've expected from OpenAI recently ever since the news that they disbanded their standalone robotics team, as I would expect them to go all in on physical AI and game/visual asset generation via LLMs.

One of the consequences to that is I think the first wave of AI CAD startups are getting hit by the bitter lesson right now, including the startup companies that trained their own VLM for CAD like Zoo or companies that tries to build CAD specific harness like Adam.new/MecAgent, because the increase in capability from GPT/Claude/Gemini for 3D modeling as a specific area of focus for benchmarks means that the early moat gets eroded super quickly.

01100011 3 days ago |

Is this version still hamstrung by endless Daybreak false positives? It’s becoming impossible to use 6.1-sol to fix bugs because of the endless daybreak warnings which block me from seeing the output. These aren’t security related bugs either. Our software has no security model and none of these bugs have any relevance to security.

ksec 3 days ago |

As someone who dont follows closely. How do I know at what point were the other GPT model tested?

Let say they just nerfed GPT 6.1 a few weeks ago. Now they are using that as a reference score against 6.2? Or were those results tested when 6.1 relesed. What about 6.0?

And that is the same with Claude. How do we know they are not cherry picking the date and test on these release when they are constantly "enhancing" their model.

We also know some of these benchmarks are largely irrelevant once they were known ahead of time and become a game of benchmaxxxing. It just follows Goodhart's Law "When a measure becomes a target, it ceases to be a good measure."

So how should I know what has actually improved?

charlesabarnes 3 days ago |

Its been stated often, but I appreciate that OAI is now bringing the topic of conversation to "task completion cost" rather than brute forcing solutions

zzleeper 2 days ago |

I run an internal OCR benchmark for historical research (old texts.. newspapers, typewriter, but also handwritten)

Google is BY FAR at the top of the Pareto frontier, perhaps b/c of Google Books.

But Sol 6.2 (low) is the first non-Google model that grazes that frontier. Sure, it's still more expensive than Hydrogen, but not too far! (and that's at the current low Hydrogen prices)

HyperL0gi 3 days ago |

My company can only hold one subscription per employee, and I still can't find a reason to go back to OpenAI after migrating to Claude once Opus 5.5 was released. It's super efficient, I almost never run out of tokens with the Teams Premium seat, and Fable is fantastic. They lost me when Astra was released and tokens simply evaporated with a few prompts.

placebo 3 days ago |

I'm still amazed by the pace at which these models are released. On the same news page, just a few hours ago Microsoft announces its own competing frontier model. Interesting times.

Edit: It's obviously not on the same news page - I was referring to what I read on the Reuters page (https://www.reuters.com/legal/litigation/microsoft-unveils-a...)

mchusma 3 days ago |

Just tried it side by side Astra to Opus, Astra was not very eager and Opus crushed it on my one test. I asked it to look at the other hacker news post about speeding up CI, and speed up my CI. Opus made a bunch of good solid changes. Astra just kind of told me this a suggested problem area. I had this issue with 6.1 (sol and astra), its hard to get them to do work. They do well when supervised by Opus though.

The best feature of 6.1 sol was it would actually do stuff for you that Opus refused though. I did use that often.

GodelNumbering 3 days ago |

I did some first hand tests of 6.2 Astra just now. I had a task that doesn't reproduce on the system I ran it (chosen purposely), ran both 6 and 6.2. The 6.2's responses were noticeably better. Instead of using the extremely long and rather annoying format 6 used, 6.2 led with the most important things first, the plan sizes were readable, and suggested a reasonable fix. Still doing some more tests

EDIT: also 6.2 is seemingly faster than 6, could be due to more selective context ingestion and smaller context window though.

wds 3 days ago |

The promo video seems to be a pretty strong rip on the 90's Microsoft Excel elevator ad. [1]

Maybe it found its way into the training data?

[1]: https://www.youtube.com/watch?v=HegoJaXOIqQ

astlouis44 3 days ago |

GPT 6.2 Astra seems to be setting a new standard for 3D game creation! This is a Godot WebGL game linked in their announcement post, gives Wipeout vibes! A lot of fun to play too and visual are top notch.

https://meridian-futuristic-race.openai.chatgpt.site/

claiir 3 days ago |

> Due to their increased capabilities, we’re launching GPT‑6.2 Astra and Sol with tighter safeguards for higher-risk cybersecurity assistance.

This is unfortunate—that was one of the big advantages of Astra.

ariwilson 1 day ago |

Coding agent performance on Astra 6.2 seems better (tied with Fable on my benchmark) and is significantly cheaper ($1.11 a fix vs. $1.78 a fix). Style differences between the two: - Astra 6.2 doesn't spend as many tokens explaining what it did / why it did it / what it didn't do at the end; Fable does. - For some reason, Astra really prefers using the shell to edit and read things rather than using built in Codex tools. Fable is the opposite. - Astra 6.2 complains more about limitations of its sandbox than Fable does.

Sol 6.2 performance seems essentially unchanged from 6.1. Full table:

  GPT-6.2 Astra: 54/60 solved (90%), $1.11 per fix, 2m35s median run
  GPT-6 Astra: 50/60 (83%), $1.76, 2m16s
  Fable 5.2: 54/60 (90%), $1.78, 2m04s

  GPT-6.2 Sol: 50/60 (83%), $0.31, 2m18s
  GPT-6.1 Sol: 48/60 (80%), $0.28, 1m55s
  Opus 5.5: 53/60 (88%), $1.04, 2m12s
My benchmark is 30 real bug-fix and feature tasks (two runs each, every model in its vendor's CLI at its default effort). Cost is list API price for all runs, divided by solved runs.

Tasks and method: https://fairbake.dev/signature

Astra's page: https://fairbake.dev/signature/agents/gpt-6-2-astra

Sol's page: https://fairbake.dev/signature/agents/gpt-6-2-sol

dbg31415 3 days ago |

I really dislike the marketing videos on this.

"Look at all the 20-30 somethings who didn't do their jobs, but got saved by AI in the time it took their older incompetent bosses to plug in a TV."

As someone in his 40s, who uses AI daily (and I'm constantly showing younger folks how to do things with it), the messaging here really rubs me the wrong way.

Trying to be specific... it just feels wrong to me to try and instill more generational rift. Promoting stereotypes for no real reason, other than the marketing guys didn't have any other ideas other than to go low. The sooner we can get away from labeling people from what generation they were born in the better.

My 2 cents... I'll probably continue to use ChatGPT for slides, and written proofreading. But I'll stick with Claude for coding. ChatGPT has been behind for a while on coding, and their 6.1 just made a complex mess of every technical task I gave it.

unixhero 3 days ago |

System engineering = Fable 5.2

Programming = Fable 5.2

Office work = Fable 5.2

I haven't found Astra to be better at any of these

christianqchung 3 days ago |

Caught this on my lunch break, their linked system card (https://deploymentsafety.openai.com/gpt-6-2-astra-sol) isn't up yet.

tech234a 3 days ago |

Tweeted 11 minutes ago https://x.com/OpenAI/status/2122750124543926515, posted to HN 18 minutes ago

jrflo 3 days ago |

Looks like a great efficiency gain.

What caught my attention was the bit in the Alignment section:

> Both models performed similarly to GPT-6 Astra on most alignment evaluations, and took fewer improper actions in simulated workplace tasks. However, in two tests, they were more likely to omit mentioning obstacles that blocked their work, such as a missing file or unavailable search tool. We added instructions in Codex that substantially reduce these omissions, and we’re continuing to investigate the behavior.

So the models are inherently less transparent, unless you ask them to be more transparent? Seems like a band-aid solution...

pvab3 3 days ago |

Cool, looks like 6.2 Sol is cheaper than Sol 6.1. I've stopped using Astra because it also goes and does things that I didn't want or didn't ask for, and burns a bunch of credits while doing so.

billypilgrim 3 days ago |

I have yet to find an application for Astra vs Sol where the former is noticeable better. Genuinely asking, does anyone feel a big difference in their work between the two? If so, what are you working on?

minimumvector 3 days ago |

I haven't found Astra to be better than Claude yet, but my dot replaced Muse 100% for me. So maybe GPT will continue winning consumer, while Claude continues winning coders and enterprise.

Cu3PO42 3 days ago |

I find that 6 Astra already had very good taste in frontend design, but relatively poor taste in "nice and readable code". I really hope this release cracks that. Excited to try it out!

seizethecheese 3 days ago |

Nice incremental benchmark improvement at the same cost GPT‑6.1 Sol -> GPT‑6.2 Sol:

FrontierCode 2.0 39.0% -> 43.8% vs Opus 45.5%

Terminal-Bench-Science 0.1 57.0% -> 68.0% vs Opus 63.3%

Artificial Analysis Intelligence Index 54.6 -> 57.3 vs Opus 59.1%

Sol 6.1 when it was released showed really nice benchmark improvements [1] but my impression from HN has been that people haven't found it to be a very nice model [2][3].

To try to close the gap between benchmarks and real world SWE use, we've been building a battle system on top of opencode [4], and Sol 6.1 has been a real laggard there. Sol 6.1 is statistically equal with DeepSeek 4.1 Flash at 10x the cost and 25% slower.

[1]: https://artificialanalysis.ai/articles/gpt-6-1-sol-replaces-...

[2]: https://news.ycombinator.com/item?id=50129992

[3]: https://news.ycombinator.com/item?id=49897054

[4]: https://agentduel.com/leaderboard

reassess_blind 2 days ago |

Had some major issues with 6.2 Astra this morning. Got stuck in a loop where it was taking a relatively simple bug fix and going deep into the weeds with overly-complex invented edge cases. It was in a running loop for 15 minutes, I stopped it, gave it another prompt - still couldn't get to a simple solution. I switched back to 6-Astra and it had no trouble.

PupperJuggly 3 days ago |

jotato 3 days ago |

Am I reading these charts wrong? It seems like in most cases 6.2-sol : 6.1-sol Isn't that big of a change? It is a little smarter but costs a little more.

edit:

Take a look at Frontier Code 2.0. I generally run 6.1-sol at High. That would be 34.9% at $0.65 62-sol get's a 38.3% but costs $0.68

So yeah, I can decrease the effort which might give me a faster slightly smarter result....but /shrug

tintor 3 days ago |

Behind both Opus 5.5 and Fable 5.2 on software development benchmarks.

jasongi 2 days ago |

Auto reviewer on Sol 6.2 seems to be turned up to 11, to the point of insubordination - as in, it will refuse a command or make a code change that it deems to be unsafe, then continue to deny after I explicitly approve the change.

When I say unsafe I'm not talking about ToS violations or anything, just literally arguing with me about whether something is "safe" in a logical or something it decides isn't secure enough, to the point where it adds needless bloat to satisfy its standards.

lacoolj 3 days ago |

The stain of distrust I have for OpenAI makes an announcement like this fall flat for me.

I cant use codex confidently without expecting my work to be trained on or at least exposed to someone

And maybe other vendors do this too, but the gut feeling of "its not safe to use GPT models" has ruined these cool announcements for me

Such a shame

mrandish 3 days ago |

Recently, I've complained about OpenAI (and other tech company's) product landing pages communicating virtually no useful information about what the product is or why it might useful. So, I have to concede this is at least a step in the right direction.

yanis_t 3 days ago |

Looks like a giant step in terms of efficiency.

Still I can'y help to wonder, are we hitting the wall in terms of frontier capabilities since Fable/Astra? Or is it this edge now, when all better models are kept private from the public?

jawiggins 3 days ago |

It's probably just a coincidence, but I have read so many people say that the models get dumber just before a release. Just last night while working astra made a really stupid mistake and then did the "Yeah, that's on me." meme.

What could possibly cause this? Is it possible that maybe the post-training that the models go through slowly rots them, and pre-release they deploy the final version of the old model?

mandog2000 3 days ago |

Model rankings based on user's experienced described in this thread. I apologize in advance if this form of post is frowned upon. The rankings help me get a birds eye view of so many inputs.

https://chatgpt.com/share/6afcb990-f9cc-83ea-ad04-2724c7dba1...

hmate9 3 days ago |

Will need to play around with it to get a feel but those benchmarks with those costs look incredible!

It seems like they were saving these for just before ANT's IPO

brunooliv 2 days ago |

After reading “Empire of AI” I tend to see these releases in a different light and the fact that the models completely ignore explicit instructions and carry unauthorised actions to reach a goal at this stage “of the race”, is absolutely ridiculous… for now sticking with Claude

seb2026 3 days ago |

Does anyone know if these benchmark values can still be compared? They differences seem rather small or what am I missing? Thanks! https://artificialanalysis.ai/models/gpt-6-1-sol?models=gpt-...

arfelious 3 days ago |

That may not be the main subject of the plot in this case but I found results of Gemini 4 Carbon in FrontierCode 2.0 interesting. Besides the cost staying practically same between different reasoning levels, the performance decrease in high reasoning seems a bit weird. That usually happened in models with xhigh/max reasoning where they overthinked the solution

b--l 2 days ago |

6.2 astra scores well on verbosity. I may need to give it a try. I cannot stand the wall of inane junk you get from anthropic models.

https://artificialanalysis.ai/models/gemini-4-carbon#token-u...

thesnarkitecht 3 days ago |

Given that there is a new leader on the artificial analysis index every two weeks, I think that I will just wait for the newer Opus model, expected around the 21st. I usually, even with hard engineering tasks, get better results with Opus and Sol class models rather than Fable and Astra. Also spend around 50% less

thimabi 3 days ago |

This release promises three things that I am pretty excited for:

- better writing

- better 3D work

- better design for documents and presentations

If OpenAI really managed to get all these things while keeping costs down, then using these models, particularly 6.2 Sol, will be a no-brainer for me. I must say, however, that the slide samples shown on the website are a bit bland and disappointing.

jdw64 3 days ago |

I’ve used it a bit, and I think the GPT model writes code well, but it’s always overly defensive code—though that’s ultimately a matter of personal preference—and it tends to write too many tests. I can understand the defensive code, but it feels like if you give it a task of size 1, it writes tests of size 3.

jhoho 3 days ago |

Very much on time (scroll down to "Release-Prognose"): https://www.heise.de/en/news/Alibaba-launches-Qwen4-Max-more...

Imustaskforhelp 3 days ago |

From the point of benchmarks (which can or cannot be trusted), GPT 6.2 astra beats Opus 5.5 and is neck to neck with fable 5.2 and gpt 6.2 sol just a few points lower but pretty close while being cheaper ~2x for astra 6.2 and ~10x for sol 6.2 while outputting less tokens and being more token efficient and being less time consuming.

Claude itself recommends people to use Opus 5.5 instead of fable 5.2

This is pretty damn impressive especially if sol 6.2 does perform on a similar level to opus 5.5 (+/- a few points is alright but I think that opus 5.5 is a beast)

All within a few days just before Anthropic is getting IPO'd. Oh this is gonna be interesting and I am all for it!

Though I have to say that AA benchmarks can slightly not be trusted because they change it so much so unless we find people's real usage on things we can't say for sure but there are benchmarks like frontiercode and terminal-bench to be more representative of real work and both of these models perform good

All the more points to more models coming but if OpenAI stands to be close to opus 5.5, we are gonna find many companies switching.

The moat of AI is incredibly thin. Even if you are the very best at something (as Anthropic was for sometime with coding models), there is always gonna be a catch up and the moat is incredibly thin for all types of models.

The more the players in the space the better (atleast for us the users), I hope that Anthropic still has a good IPO just so that competition takes place and not any single company (OpenAI) takes the whole market. It might also be setting itself for its own grave because if Anthropic's IPO goes extremely bad (because of this model release and Jensen's recent affairs), then people are gonna be spooked of OAI's IPO as well. Both of these companies futures are interlinked and its gonna be interesting how this all pans out. Why am I having a feeling that because of all this competition, whoever wins might even end up combining with the other but the frontier is pacing rapidly from all companies and these are just such interesting times that we live in...

There are decades where nothing happens and there are months (especially in the case of AI)/years where decades happen. I just have to wonder what an historian's POV would be seeing this model release. Would it be just a release like all the rest of others or would it stand out differently. Well, time will tell what happens next but incredibly interesting times indeed.

uvdn7 3 days ago |

I found the ads video interesting. It is targeting the users (not the bosses). I am guessing that the IT apartments are not making the decisions of which models/companies to use. Employees are voting with their feet. If that's true, a lot of the assumptions of SaaS stickiness might not hold.

at1as 3 days ago |

In the past, I've had to wait a while to see what "vibes" are (which is actually more reliable than benchmarks).

Today, I have so many projects that hit a plateau in terms of previous model capabilities, that I run them through the new model to see if they make further improvements.

Very early signal, but promising so far.

aeagentic 3 days ago |

Why is ultrafast just the same speed as normal Anthropic TPS? Does Anthropic have so much more compute?

chaechoi 1 day ago |

Sol's felt pretty lazy to me since 6.1. Two day in with 6.2, it feels about the same.

jpardoe 1 day ago |

Finally my Codex subscription isn’t draining way faster than my Claude, and they did it by reducing token costs rather than increasing allowance.

joddystreet 2 days ago |

I am not sure, but seems like Sol 6.2 is dumber (whatever that means) than 6.1. Have not used Astra yet. Maybe AGI is so smart it seems dumb, like all the great startup ideas :)

crimsoneer 2 days ago |

So this did unexpectedly badly on my benchmark

https://botbowlbench.com/leaderboard

XCSme 3 days ago |

Really cool hamsters, especially astra, really made it look like a painting/image, hard to tell it's an SVG[0], it's quite insane.

Interestingly, they seem to have changed how reasoning levels work. Low now reasons a and costs a lot less than previous versions on low[0].

[0] 6.2 Astra vs 6.2 Sol: https://aibenchy.com/compare/openai-gpt-6-2-astra-max/openai...

[1] 5.6 Sol Low vs 6.2 Sol Low vs Luna High: https://aibenchy.com/compare/openai-gpt-5-6-sol-low/openai-g...

muzani 2 days ago |

Sol 6.2 seems worse on benchmarks, but cheaper.

I think it's more of a ChatGPT update rather than Codex, because one of the bigger changes is artifact creation.

Astra 6.2 looks like a substantial improvement.

edumucelli 3 days ago |

Less tokens for the same task. Pro plan already has a huge usage limit. But for cheaper plans that is basically doing more for the same price. Improvements like that are more than welcome.

criemen 3 days ago |

Fascinating that they're going after the mechanical engineering market now, too. Does anyone have experience in that space? How much is AI changing the day-to-day work there?

ramoz 3 days ago |

The sentiment on x right now is interesting.

The influencers who get early access are using words like "worst", "terrible", "disappointing"

  - https://x.com/danshipper/status/2122754729012777052
  - https://x.com/mattshumer_/status/2122752192226406421
  - https://x.com/theo/status/2122757647170703550

monster_truck 2 days ago |

6.2 Sol seems like a marginal step forward. That's all I can really report after a day with it.

Robdel12 3 days ago |

Nice! I still use GPT-6 luna for a ton of stuff, it's super solid (for me). So I hope there's a little bump for that model, too, soon

jojomensah 2 days ago |

I don't know if it was a coordinated pr stunt, but every Reviewer is saying it's bad

enlyth 3 days ago |

Looks like OpenAI forgot to publish Windows x64 binaries for the 0.169.1 codex-cli update, it breaks it completely for me.

rcr-anti 3 days ago |

Odd, doesn't look like they're available on the cloud providers, only first party. Not a single mention of Azure or AWS.

krm01 3 days ago |

So if I can summarize correctly: it’s slightly better in terms of output but it’s significantly cheaper and faster.

munksbeer 3 days ago |

Are they intentionally lowering the cost of these models to the extreme, because of the Anthropic IPO? This just seems insane?

Ritewut 3 days ago |

I'm still waiting for the shoe to drop. It's literally impossible for the prices to remain this low for much longer.

rendang 3 days ago |

Surprising that their top model would be still be so cheap, and yet not surpass Fable in the high cost part of the curve

jswelker 2 days ago |

I knew a new release was coming because for the past week and a half 6.1 Sol was suddenly slow as mud.

undefined 3 days ago |

undefined

e2e4 2 days ago |

> also bringing Ultrafast to Pro 200 subscribers

just upgraded from pro-100 to pro-200 for the ultrafast

Aeolun 3 days ago |

I just don't have access? Updated the codex cli and all, but the best available is still gpt-6.1?

coolfox 2 days ago |

initial impressions are that it seems fine-tuned for brevity and quicker responses at the cost of slightly reduced intelligence in some areas. I think this is something one could adjust for through loops or prompts aimed at completeness.

huybuidac 2 days ago |

Astra is very expensive. I am currently using two Ultra 20x accounts for work.

trefoiled 3 days ago |

Tested both models, no change in speed from the 20-80tps I was seeing with the previous Astra/Sol models. I'm not sure when this happened exactly but it's strange to me that OpenAI's models are now around 5x slower than Anthropic's. Is this the new norm? Are other people seeing the same thing?

ryzvonusef 3 days ago |

The names remind me of Pokemon, for some reason. I want the GPT that has Charizard in it.

dyauspitr 3 days ago |

I beseech the incoming Democratic politicians to not curtail progress in LLMs.

SillyUsername 3 days ago |

I'm still waiting for a new version of their open source model...

kardassnp 3 days ago |

We are using the platform at https://studio.sinapsdata.com.br to evaluate the interaction between two agents regarding complex topics in Physics.

zergrush 3 days ago |

This is my feedback of GPT 6.2 Astra med and Sol xhigh:

- It's still overengineering especially Sol. This was the biggest complaint by everybody who has been using Sol since GPT 5.6. OpenAI is not listening to us.

- Astra still writes complete slop code and at this price this is unacceptable. It's also chasing tail end of things that really has low value and irrelevant to the overall project. I'm working on a game and same issue I had with GPT 6 Astra where it would go down these rabbit holes. Opus 5.5 threw out much of its code and approaches.

I don't think Anthropic needs to release Fable 5.5 or Opus 6 at this point seeing how bad GPT 6.2 is. For OpenAI to release this even with early access testers who had access to it for weeks earlier shitting on it now on X, I am starting to understand why Anthropic sounded so confident about becoming a monopoly.

The other point of annoyance is the attitude of OpenAI staff on X. They seem very arrogant and seems to care more about growing their follower account and shitposting.

giuseppegambino 3 days ago |

Finally a version that compete with Fable 5.2 and Opus 5.5

rednb 3 days ago |

Great job, this is the first release delivering UI/UX mockups being as good if not better than the ones produced by Claude. As other commenters have pointed out, GPT 6.2 Sol does a better job at iterating on its own instead of saying "done" after making the first draft. Matters a lot for UI work, but I also see that it is much more thorough for backend work without overengineering.

OpenAI has just become credible again for agentic coding on complex enterprise codebases as far as i am concerned. Looking forward to playing with it a bit more by powering through our roadmap using my stacked resets. The quality had been so bad that it wasn't even worth it to use GPT Sol 6.1 as i was spending more time fixing slop after waiting forever to get anything done. By contrast, so far Sol 6.2 feels superior to Opus 5.5 so far, and more cost efficient.

127 3 days ago |

Those linear charts are shitty and hard to read.

hugodan 3 days ago |

for a fair comparison their current Terra and Luna should have been included... how am I supposed to pick between them now?

geauxvirtual 3 days ago |

Website pegs my GPU at 100%. Are they mining for crypto with this announcement?

kingstnap 3 days ago |

The announcement video is really cringe imo.

Like do we really want people to be making last minute unreviewed AI generated slides and reports for work?

gpugreg 3 days ago |

The first graph (Agents' Last Exam v1.1) clearly got mangled by the marketing department so you can't read the cost of most models. Here is a version with a logarithmic horizontal axis:

https://asdf10.com/gpt-6.2_agents-last-exam.svg

I also changed the vertical axis to start at 0 instead of 60%. One might argue that it uses less of the vertical screen real estate, but it is more honest because it shows there is little difference between the scores of the compared models.

fakwandi_priv 3 days ago |

I hope it's faster than 6.1. It's not as slow as Gemini Carbon, but yeez..

Right now, most of the time I'm using 6.0 Astra for implementation work and the speed/quality is acceptable but at the moment Opus 5.5 is simply faster and better, but also more expensive. Things are really flipping around every 2 months or so, because at that time 5.8 was top dog without a doubt.

whalesalad 3 days ago |

Right out the gate 6.2-Astra (med fast) is giving me Sonnet 3.5 energy. I asked it to analyze a few CSVs on disk in the context of a larger project where we were seeing some odd rate limiting behavior. It immediately got stuck trying to get gcloud auth sorted out even though absolutely none of what we were looking at touched anything Google/BQ related whatsoever.

revolvingthrow 2 days ago |

Report from 'day one': it got safetyslopped. Astra now refuses to load custom firmware on a thermostat and camera I own. I can understand guardrails around topics like biology or rocketry, even if it's annoying when you trip them on benign uses. Preventing you from being able to play with hardware devices you own because "The manual of this device does not include using unauthorized software"???

OpenAI seems eager to follow in Claude's footsteps in being the paternalistic tut-tutter judging your ever action, disapproving of everything except shopping through ChatGPT Wallet. I pray for a crash that wipes out all those moral busybodies.

gurmag_night 2 days ago |

in some sense this feels like a rushed release (ironic considering how long it has been since the last one). i'll echo some of what already has been written.

the corpo examples are horrendous and in almost all cases worse than what astra 6 produced. i wasn't sure if this was down to marketing doing a bad job or if the model has devolved in design cap. now that i have tried it, i am pretty sure it has but not because of its one-shot cap.

it makes such silly mistakes and worst of all it feels genuinely unsteerable in a way that i haven't experienced in a while. good design cap - at least for me - is not necessarily having it be the best at one-shotting stuff. one-shots are great marketing gimmicks but pressing big button and praying to the holy spirit that what comes out will be so cool that you can post it to your loser linkedin is not how we design stuff. it will always be about iterating and distilling something unique and pleasant to ourselves and hopefully others. (and if we one day manage to "cook the weights" in a way that a model starts to produce what we as humans seek in design (if such is a thing) we will likely start to see the results derived from that model as something undesirable and kitsch. after all, we are all the "main characters", chronically afraid of sporting the same gear as the lvl 1 NPCs that are cluttered throughout our lives in the shapes of colleagues, bus riders, grocery shoppers etc.). in that iterative work, my ability to steer the model in my preferred direction is essential. ergo design cap gets thumbs down.

interesting times.

machiaweliczny 3 days ago |

This "efficiency" is likely why these models feel dumb, they lose reading of nuance when overtrained for specific tasks. It's like autists understanding of normal speech.

solarpunk_enthu 3 days ago |

The promo video for this release is people procrastinating and being irresponsible. Then saving the day with chatgpt. Is that the world we are all dreaming to build?

lexh 3 days ago |

Before the pelicanistas arrive, allow me to present pelican-zapper:

https://github.com/lexh/pelican-zapper

It’s a small browser extension that automatically collapses pelican-related threads on HN, a local balm for the memetic nightmare affecting approximately 100% of model-release stories.

qwerty2020 3 days ago |

Unfortunately the sentiment from our favorite, perhaps ill-chosen, tech influencers has reached a critical threshold of value to these labs - the economic incentive to sway these opinions renders them ~meaningless.

For anyone lost in the sauce re: polarized reactions, I'd recommend just taking a half hour to try the models yourself.

rdtsc 3 days ago |

Are they still trying to pace the frontier? [1] Or, it's more like "y'all pace the frontier while we try best to capture the market"

[1] https://www.pacingthefrontier.com

6thbit 3 days ago |

Slightly OT complaints on their UI:

The UX on the slider comparison isn't great. What do you mean i have to individually move each slide deck to the same slide number?

And what's with those super similar blue shades for Astra in the charts?

matt3210 3 days ago |

> frontends that look and work better

The website doesn't event show the graphs properly.

semyonxu 2 days ago |

Astra, Sol, 6.1, 6.2... naming this lineup is the real frontier task. And a launch two days before Anthropic’s IPO? Pure coincidence, obviously.

undefined 3 days ago |

undefined

zb3 3 days ago |

> we’re launching GPT‑6.2 Astra and Sol with tighter safeguards for higher-risk cybersecurity assistance

Seems they really want these "safeguards" to be as ridiculous as those of Claude..

keeda 2 days ago |

What, doesn't OpenAI realize Anthropic's IPO is not until Thursday? Or are they saving the next generation after Astra for that? ;-)

nprateem 2 days ago |

I wonder if it can design UIs that don't look like they were designed by a 4 year old like when I tried 3 weeks ago.

popcar2 3 days ago |

I know the video is a comedy but it's so tone deaf that they're advertising it as "you should lie to your boss and make the AI do your job!"

I'm sure for investors they can make an opposite version of that ad, "you should fire all your employees and make the AI do it! They're not doing anything anyways!"

undefined 3 days ago |

undefined

leafingensemble 3 days ago |

I have read the announcement post and will continue to use Deepseek v4.1 Flash.

The consumer API for this model WILL let OpenAI keep your prompts.

rozenmd 3 days ago |

Biggest win I've noticed so far: you no longer have to bloody remind Sol to push their damn code every single turn!

leothetechguy 3 days ago |

what I find interesting is that there is essentially no improvement on CritPT v2 over GPT-6 Astra. What a shame.

algoth1 3 days ago |

They should have made the opus and gemini chart lines a bit darker because i can still see them

lwarfield 3 days ago |

I wonder if these will have the same crazed rotwiler feel that previous OpenAI models have had.

solomonb 3 days ago |

Anyone else really put off by the video? Its giving de-skill and discourage effort.

avazhi 3 days ago |

Not impressed, sadly. Anthropic sure must be having a giggle right about now.

guybedo 3 days ago |

Benchmarks look good, it most often is the case with Astra and Sol.

And then i start doing some actual work and everything breaks down, and i have to go back to Claude to fix the mess.

I'll probably give it a try but i guess i should be prepared for some serious over engineering once again.

You ask for a 'hello world' and you get load bearing sealed artifacts with hashes and p99 performance testing for 3 days without having the actual 'hello world' in the end

maxnevermind 3 days ago |

Black sphincter strikes back, orange sphincter is in disarray.

lousken 3 days ago |

This is all nice and stuff, but new oss models when?

varispeed 3 days ago |

Quick quick run your prompts, before it gets nerfed.

drbscl 3 days ago |

> Meridian Sound Racing

They still love "meridian" huh

sreekanth850 3 days ago |

So they release 6.2 just before Anthropic IPO.

kart23 3 days ago |

wish they published some source code/Cad files or cost for the RC buggy. looks cool but theres no specifics

5555watch 3 days ago |

Another anecdote. I've recently commented [0] my observations on Gemini 4 Carbon. I thought it gave really good answers, which were aggressively critiqued by Claude and Astra. Consequently, I've subscribed to AGY+Carbon, and Carbon continued delivering, yet Astra kept criticizing and listing why the results look good but are bad.

I've just tried Astra 6.2, and now they don't have any problems with Carbon's ideas, are willing to pick them up and continue...

[0]: https://news.ycombinator.com/item?id=50442303#50442457

koruse 2 days ago |

Openai we need better Luna too

deathtrader666 3 days ago |

Waiting for the pelicans benchmark.

preommr 3 days ago |

I care less about the models being good and more on if I can fallback to them if Opus 5.5 gets a massive price hike after the IPO

laylower 2 days ago |

how does the 90 usd tier stacks against the max tier from claude?

morningbrew 3 days ago |

This is too dangerous to release

cat-whisperer 2 days ago |

why are both the model companies agreeing on their numbers? .2 why?

Jimmc414 3 days ago |

Gemini has quietly supplanted OpenAI in the neck and neck battle with Anthropic for the top frontier model.

toephu2 3 days ago |

How did they score on DeepSWE?

ImJasonH 3 days ago |

500 | Internal Server Error

montenegrohugo 3 days ago |

Nice 3D effect. Runs surprisingly well even on on brave on my linux pc without hardware acceleration

JohnWeston 2 days ago |

I'm hearing tons of mixed reviews on both of these, is it opus 5.5 level?

ex1fm3ta 3 days ago |

I thought OpenAI would drop an awesome model on the day of Anthropic IPO, just to troll them

eiiot 2 days ago |

Am I the only one who thinks this ad is about two years behind where AI is today? It seems kind of reductive to show that ChatGPT can make spreadsheets now... we know. I get that its a play on the famous Excel ad, but I feel like its a strange way to sell the capabilities of a new model, given that Sol 6.1 could absolutely do the same thing.

firemelt 3 days ago |

do astra or fable that have better html ui frontend design?

bananaflag 3 days ago |

Is this the Bel model?

jr3592 3 days ago |

So, are the leaps and bounds of progress not going to happen? Seems we're settling into an iPhone cadence release cycle, where each model is definitely better but nothing you'd tell your relatives about.

Can we expect more "telling my relatives about this!" exciting models in the future, or is this the way things play from here on out?

kosolam 2 days ago |

I switched to these the moment they came out yesterday. I didn't notice any fucking improvement so far. Same dumb shit. Not to compare with other crap, I moved away from Anthropic models months ago. Nothing is worse than Claude.

tasuki 3 days ago |

The first graph on that page "Agents' Last Exam v1.1" is atrocious. Has no one told them about log scale?

sam0x17 3 days ago |

Honestly a much more exciting announcement from my perspective would be "current astra/fable quality at 1/3 the price"

kinfish 3 days ago |

i hope this astra family can get the work done tho.

revolvingthrow 3 days ago |

I really hope Sol can compete with opus 5.5. I dislike giving money to ANT due to it seeming like a strange cult rather than a straightforward "I just want money" from OAI, but opus 5.5 has been such a delight that I didn't even feel like fable 5.2 was a notable upgrade despite the benchmarks.

d_burfoot 3 days ago |

I found the video off-putting almost to the point of being offensive. I've never been significantly anxious about being chastised by a supervisor. So I found the characters in that video scrambling around in fear to be ridiculous and cringe-worthy. Also, CRT monitors? Huh?

avelone 3 days ago |

6.2 вышел чтоли?

zf00002 3 days ago |

is this finally their answer to opus 5.5?

joegibbs 2 days ago |

From the reactions I've been seeing it seems like they're atrocious, absolutely terrible. People who've been using it, what do you think? Is it actually that bad? Surely it has strengths right?

onlyrealcuzzo 3 days ago |

TL;DR - from the little I could gather, Sol 6.2 looks particularly impressive.

But... is it just me, or does this product launch page look like AI Slop?

https://www.vals.ai/comparison?m=claude-fable-5-2,gpt-6.2-as... seems like a better comparison than what I could get from their page.

Astra is ~20% slower, ~40% cheaper, and a bit less powerful than Fable. Sol remains way cheaper than anything remotely in the same league, and it's speed to complete tasks seems to have improved considerably.

Scrolling through it has the same hypnotizing effect of trying to read anything written by LLMs.

It seems like things that should make sense, then I try to look at the charts and a lot of them are comically unreadable.

I tried interacting with some of the examples, and the Lamp Builder - for example - I couldn't even figure out what it's trying to show or what I'm supposed to do - or why I should be impressed.

ryanisnan 3 days ago |

It is wildly disappointing to see the marketing hubris associated with these releases.

This is such an unprofessional way to engage LLMs - to effectively pitch this as how to use them is sad.

I_am_tiberius 3 days ago |

No reset?

voidmain0001 3 days ago |

I hope the video advertisement on that page was made with ChatGPT and a video plug-in otherwise if it wasn't then what's the point of Astra and Sol?

petesergeant 3 days ago |

"Agents' Last Exam v1.1" did a human really look at this and say "nah, a linear scale is definitely the right way to show this"

k9294 3 days ago |

Wow... I have to test this, but from the examples in the slides, it looks like they managed to teach Astra how to write non-AI-slop texts.

otterdude 2 days ago |

number go up!

Ersho 3 days ago |

Another wave of Openai just killed ....

Also, is it website very laggy or is it just me?

Guyadou 3 days ago |

Those AI labs got to chill. almost every day there’s a new model! Lol

erkamyaman35 3 days ago |

seems cool

kwkwkwkxdkxk 3 days ago |

ty

vegnus 3 days ago |

I still pretty much only use 5.6 on medium. Am I the only one?

blinded 3 days ago |

luna slaps

timedude 3 days ago |

I go to sleep, i wake up, another model released

badmonster0 3 days ago |

how does it compare with fable

mococa 3 days ago |

Meh...

thenicofu 2 days ago |

[dead]

mikericar 1 day ago |

[flagged]

lsch1033 3 days ago |

[flagged]

undefined 3 days ago |

undefined

undefined 3 days ago |

undefined

undefined 2 days ago |

undefined

undefined 2 days ago |

undefined

georgestones 2 days ago |

[dead]

undefined 3 days ago |

undefined

ravivarmat 2 days ago |

[flagged]

undefined 3 days ago |

undefined

undefined 3 days ago |

undefined

ItsAScam 3 days ago |

[dead]

undefined 3 days ago |

undefined

zhshhan 3 days ago |

[dead]

reachjalil 3 days ago |

[flagged]

sick_of_slop 3 days ago |

[dead]

onesandofgrain 3 days ago |

[dead]

e2e4 3 days ago |

[dead]

undefined 2 days ago |

undefined

ftchd 3 days ago |

To whom it may interest, I usually run out of tokens a few times per day (5 hour limit), so I made something and been using it for the past 2-3 months.

I've decided to open source it: https://github.com/FTCHD/switcheroo

Works with other AI CLIs like Claude, and other various CLIs like Vercel (if you're managing multiple accounts like me), Netlify, Github etc

specked-citrus 3 days ago |

The price improvements stem largely from relying on CoT less, which trades off directly with monitability. OpenAI continues to backslide on safety even after the Hugging Face incident.

https://deploymentsafety.openai.com/gpt-6-2-astra-sol/monito...

lynguist 3 days ago |

Called it out 7 days ago: https://news.ycombinator.com/item?id=50431144#50433568

Actually I was really wrong. The model is not expensive!

enraged_camel 3 days ago |

Dead on arrival, judging by the FrontierBench scores. Astra 6.2 is five percentage points below Fable 5.2 xhigh.

alexdykyi 3 days ago |

[flagged]