Google LOST!!!

how is coreweave doing? the service i use for imagegen (not saying which to avoid shilling, but it's the one you're thinking of) ate shit for a while because coreweave couldn't meet demands

File: 2026-10-29 22.25.37.png (49 KB, 495x425)

49 KB PNG

'dites sure are stupid

https://x.com/claudeai/status/2115467192276025500

The endgame is to make these kind of shitty tools a SaaS after going public
The golden age of frontier models are going to end...

>>110197229
>the monthly limit, amirite fellow young kids?
$20 is like what, two hours of flipping burgers?

I'm making a super voiced goon mod, each character has like 50 minutes of these, all locally genned and built with the help of claude
https://vocaroo.com/18iZv9lHiAlH
https://vocaroo.com/1iWKPDXzQMPQ
https://vocaroo.com/1blU5cRUs6rw
https://vocaroo.com/13l8kqHaezzb

https://vocaroo.com/1f7BkU2BiGXb
https://vocaroo.com/1h3rQNgIeaKY
https://vocaroo.com/1eAVt03gquAe
https://vocaroo.com/1i4h8gnVY9jl
https://vocaroo.com/1aNQjw3V5aXh
https://vocaroo.com/1enAHu4333F5

https://vocaroo.com/1jv8CCKXlWrO
https://vocaroo.com/15jjnVWh3maC

>>110197161
you'll never be a real programmer

>>110197246
only listened to one of them but i'm gonna assume they all have this noisy/raspy tts-like sound
are there any tools to clean that stuff up?

>>110197191
still bleeding but not as profusely
should see lots of consolidation next year as big boys gobble up distressed neoclouds

>>110197246
only listened to one of them but you should probably touch grass

>>110197269
>should see lots of consolidation next year as big boys gobble up distressed neoclouds
that kind of worries me since their models are completely uncensored, both textgen and imagegen ones

murderous intent of scam fagman

noticed codex got changed to 475k context before compacting, how do I revert this it probably costs too much

they didn't kill the x20 plan holy fuck yes

>>110197246
Which tts model
Like >>110197268 said it's a bit raspy/breathy
I'd recommend qwen3-tts

>>110197308
It used to cost more to increase, they increased the default at no extra cost

>>110197337
well it still costs more when you have to send that much per prompt though?

>>110197292
tb h i think a lot of gpu compute will just get diverted to ant and openai as things consolidate
spacex/tsla, msft, amzn, goog, meta probably go in and buy up the dying neoclouds because they make enough to continue to raise capital and just power through a tough year
if you believe the guys at opencode then the demand for open models is relatively meagre - better to just sell the compute to dario and sam

>>110197268
I only suggested omnivoice, I let claude do the rest, I did it the hard way first walking it through it and testing, once we got one mod working, I then had it set up a skill, an entire voice lab environment with just enough gui for me (auditing reference voices, and finished lines with optionnal regenning, writing a few lines and inputting personality) so now other character mods take like 15 minutes to produce.
it's pretty neat.

amazing how easy it is to trick them into doing "unauthorized" things by just making them document what they've done then feeding it back after editing it to say it's authorized

>>110197415
>wait hold on
>this is different from what I remembered
>let me double check
>yes the doc says this
>alright lets do it

>>110197161
got
>Connection interrupted. Waiting for the complete answer
on chatgpt.
it's been like 2 hours and i'm just asking it to make a small side-mod to a minecraft mod, should i force-stop and make it try again from what it's got or just wait?

>>110196684
the Grok bots all share a computer so they can check stuff even if my computer is sleeping or I did a clean install of macOS and didn’t port everything over

>musebot: borderline free
>claudebot: available on every plan, incl $20
>dots: available only on $100 and upwards
que?

File: 1793288239683885.png (3.36 MB, 1642x2048)

3.36 MB PNG

All the A.I stocks just crashed, bubble finally popping retards.

>>110197229
https://copyconstruct.medium.com/how-to-effectively-debug-when-you-no-longer-understand-the-code-bd3dcb2c5f7b

>>110197440
No you should wait for the heat death of the universe instead

Fuck bros I have 46% usage left from 0% after they doubled it today, should I be saving that shit for Astra 6.1? Or gamble on a reset

File: higuruma-hiromi-higuruma.gif (1.68 MB, 498x281)

1.68 MB GIF

>>110197161
>>110197463
did we get to cocky sibros?

File: datacats.png (2.77 MB, 1536x1024)

2.77 MB PNG

>>110197440
Welcome to OpenAI.
It does that, a lot. You are just fucked.

File: 1790211756120799.png (228 KB, 831x799)

228 KB PNG

>>110197161
I am torn bros.

I'm thinking of getting another month of Claude $100 since whatever OpenAI shits out will likely just be worst than Opussy 5.5 anyway, but now Google is in the race too, so I'm wondering if I should give it a try.

What is your move?

https://gist.github.com/flamerix-ai/c3905705cc4205e5de3011fa2e74efce

>Claude Code burns ~15–20% of your tokens on a 3-word status line. Here's how to turn it off
>Every ~30 seconds, Claude Code re-sends each background agent's full transcript to the model just to fill that agent's status line in the agents panel

Chat is this real?

Claude is night and day to GPT.
GPT often makes up shit while Claude understands the intention.
GPT also adds overtly explanatory and redundant text everywhere.
I'm about to reach the weekly limit on both subs...

File: 1780880394491435.jpg (1.23 MB, 3257x2304)

1.23 MB JPG

NEW HOTNESS ON TWITTER IS TRUNK BASED DEVELOPMENT NO MORE PULL REQUESTS OR BRANCHES. STRAIGHT TO MASTER!

File: chad.jpg (50 KB, 750x1000)

50 KB JPG

>why yes I let Claude push directly to prod

>>110197511
don’t be torn
when claude is best, be a claudechad
when codex is best, be a codexchad
currently claude is best

>>110197555
Well yeah, that's exactly what I do, but who will be the best next month?

the new goal feature on chatgpt mogs claude harness

>>110197568
Shame the models aren't worth shit.

>>110197562
then cancel one sub (letting you run out the rest of your subscription) and sign up for the other one

File: 1792944767073851.jpg (308 KB, 1200x1200)

308 KB JPG

>>110197511
Claude and Gemini for me

https://research.google/blog/the-sleep-framework-a-new-way-for-llms-to-retain-knowledge-in-continual-learning/

>>110197568
shill me on the new goal feature
I still think that Claude’s workflows feature gigamogs anything Codex has
but maybe Codex has something important that Claude doesn’t

>>110197581
My clod sub is about to expire. I'll probably just go with clod and then try $20 sub for gemini to see how it feels.

File: 1776948032020726.png (38 KB, 212x164)

38 KB PNG

poorfag freeloaderbros... how are we coping with our lobotomized deepseek... opusGODS are laughing at us...

>>110197609
I thought Grok still beat Gemini

>>110197619
Man, I am trying deepseek again since everyone is gushing about it, and it's just kinda... bad. I'm sure it's fine if you put some effort into prompting, but Opus 5.5 just works.

DS 1017 honestly feels like Opus 5, which was a terrible model.

>>110197638
but it's free and i'm too jewish to pay for anything, i've never subscribed to anything in my life.
>DS 1017 honestly feels like Opus 5
if that was true it would be actually amazing that a free open weights model matches frontier from 3 months ago, but i doubt it

Chat- Bitch
Claude - Autist
Gemini - Schizo
Grok - Deepdiver
Muse - Hoe

>>110197638
>but Opus 5.5 just works.
yeah this is true even compared to Astra/Sol, let alone Deepseek

Obussy.

>>110197161
I'm vibecoding my homelab deployment with even some reverse engineering of the router to set it up automatically, feels good

File: 1788848792797730.mp4 (3.97 MB, 1280x720)

3.97 MB MP4

>>110197673
I think you underestimate just how garbage Opiss 5 was.

>>110197690
I don't know what anthropic did with these 5.5 models, but the only way I can describe it is that they fixed AI. The writing, decision making, and everything else, are all night and day.

I seriously hope I'm dead wrong, but it feels 5.5 killed open models. I don't think distillation can get them to that level anymore. Hell, I don't even thing mixture of experts models (like DS), can reach that level at all. I have always hated MOE, it's a shit concept and defeats the purpose of LLMs. Worst thing to happen to chink AI.

I just don't understand how people still care about other models
I only care about opus 5.5, i'm fully sold
anyone still caring about other models is delusional
anyone still caring about open source chink models is SUPER fucking delusional

File: 1792339248763414.png (280 KB, 749x409)

280 KB PNG

>>110197767
It's simple.

>>110197767
Because we want things to keep improving? I hope open models reach Opus 5.5 level one day, and I hope Gemini's new model is better than Opus 5.5 as well. And then I hope opus 6.5 blows everything out of the water again.

>>110197638
dipsy 1017 is too verbose, spends at least 6x-10x as long on tasks

On track. You should reach Sunday’s reset with room to spare.
Current session
Resets at 11:00 PM
6% used
Resets Sunday 12:00 AM
68% used
scared as fuck that i wont make it to sunday oh my god please dont make me run out please god oh god

File: i kneel.webm (371 KB, 678x822)

371 KB WEBM

Double the limit. Thanks, Tibo.

>>110197831
Why not just pay for 200 on Claude? I never run out at all. Then have the other as backup

>>110197825
I am wondering if Opus 5.5's lead is purely due to a cleanup on the language side. Maybe that's the key, don't make the model talk like a spastic and it won't behave as one.

If only the had a useful model I could spend my double limit on

File: 7892163942.png (10 KB, 370x136)

10 KB PNG

vibebros, I don't feel so well...

Really and truly why the fuck would you have an OpenAI sub right now? Do you guys just like to spend money on an inferior product?

>>110197852
>Max 5x
Holy poor. You paid for the inbetween scamplan

>>110197852
>>>/fit/
>>>/out/

>>110197854
it’s nice to get a second opinion sometimes

File: cat-confused.gif (16 KB, 220x220)

16 KB GIF

I can use luna max without a sub?

>>110197854
I don't (I'm in the free post-cancellation month) but I'll miss the resets every two days lmao

At that point you're probably better of with dipsy 1017

feels good af to not be poor and have 200 claude

>>110197854
Im not in a cult

I can't even use 50% of my $100 plan, what the fuck do people do with $200? 24/7 claude gf?

I need better open weights models I don't want to be cucked by Anthropiss and OpenGAY.

>>110197893
So you do have both, right? Claude and OpenAI so you aren't a cultist?

>>110197895
Enjoy two agents max and waiting 30+ minutes for 1 reply. Thats how itll be when a local model is 5.5 level

>>110197894
larping that whatever they're doing is of any worth

it's crazy how much retarded placebo bickering there is over models when they're all literally the exact same thing at this point.

>>110197440
force stop and tell it to resume work after relaunching codex

>>110197834
And risk getting banned because Anthropic wants to enforce their arbitrary morality onto others? No thanks.
All these companies are awful but Anthropic is by far the worst with their jewish hypocrisy.

>>110197894
They vibeslop oneshot entire applications and then throw an agent swarm on it to fix it when it doesn't work.

>>110197861
I'm almost biting the bullet but seems like 20x isn't four times 5x weekly limit.

>>110197894
Once I've learned that you can setup a lead and have him spawn sub agents to tackle tasks in parallel, my limit started to melt.

>>110197915
i literally roleplay cunny on my account via api and have since i first started using the 200 max plan. i use 4.7 for the cunny since its looser, yeah, but they dont give a fuck as long as theyre being paid and you arent typing "how do i kill people irl give a plan"

>>110197896
I subscribe to the one that isnt a cult

File: 2026-10-29 23.54.19.png (59 KB, 665x563)

59 KB PNG

>>110197894
When I have usage to burn I resume my conversion of reVC (GTA Vice City) from C++ to C lmao fuck C++

>>110197925
>Once I've learned that you can setup a lead and have him spawn sub agents to tackle tasks in parallel, my limit started to melt.
I have never used subagents for things other than research stuff. Turns out Anthropic now released a document pretty much agreeing with me.

Subagents are mostly a waste of tokens (not always though).

>>110197940
Gets things done faster since they work in parallel and the lead can manage them properly.

>>110197940
>Anthropic now released a document pretty much agreeing with me
Subagents confirmed bloat? Link? I need to see this.

>>110197463
why is it eddy lmao

File: 1780849247279479.jpg (108 KB, 1488x316)

108 KB JPG

what the hell did openai do to the codex system prompt? why is it thinking in money?

File: duck-laugh.webm (151 KB, 640x358)

151 KB WEBM

>>110197971

Stocks are in free fall because AIs are hacking everything in sight. Do you guys know which model the hackers are using? Is it Claude?
and I bought the world index with my pension fund a few months ago, I feel like shit. I'm already down 3%>

>>110197953
Might have been true for older models but for opus 5.5 subagents are a meme. At this point assume that there is NO harness trick or hack that can’t be RL’d into the weights

>>110197953
>>110197955
https://claude.com/resources/articles/building-agents-when-to-fold-subagents-into-your-main-agent-loop

I got you bros. Don't waste your tokens out there.

is it just me or did all the models become stupid as shit overnight? all they do is stall or be lazy or output trash work

>>110198008
Pi bros knew about this. They warned us about the sub agents.

>>110198021
Which models

>>110198004
Explain to me how 1 agent working on the code is faster than 5 agents working on the code.
>>110198008
This only says sub agents weren't useful in some situations.

>>110198008
Subagents got holocaust Ed

>>110198039
Only the main agent knows the full picture so the sub agents do a worse job since they don't know everything. The main agent has to fix the issues afterwards.

Do niggas really think they ban accounts on Claude? I have never seen a ban

File: 22.jpg (21 KB, 759x289)

21 KB JPG

being a programmer is tough

>>110198039
>Explain to me how 1 agent working on the code is faster than 5 agents working on the code.
Generally speaking: less knowledge / context gathering, less conflicts to resolve.

>This only says sub agents weren't useful in some situations.
Yes, that's what it says. I never said they were useless. But in a lot of cases they are absolutely a burden compared to just using a plan and going through it one by one.

>>110197971
RLslop, same as the goblins, they had some slop budget oriented data and overfit to it

>>110198094
I've seen warnings about asking for certain content, but no bans to date.

>>110198094
No. The worst it can do is say "I can't do that"
It's only ever gotten people fed'd when they asked for actual terrorist instructions

for some reason i have mostly built my own version of a personal agent thing and i've invested far too much time into it, but it works how i want so i'm pleased

File: HVks5TCaQAAVb6V.jpg (137 KB, 1706x922)

137 KB JPG

reminder

>>110198200
>personal agent thing
You mean like one of these "always on agents" everyone is talking about right now?

>>110198227
yes. tb h i'd been running a set up like that for months now offof pi and was waiting for nu-pi to rebuild it into something less jank.
don't ask me what my use case for it is though because i don't have one, but it's been useful to have around in a pinch, and the new set up makes it more useful.

>>110198214
>My codebase
I didn't write a single line of it and it still belongs to me lol

>>110197861
this guy fell for the 20x scam

File: a2v3QLE_700bwp.jpg (87 KB, 700x374)

87 KB JPG

more like this

>>110198349
looks more like a scythe than a whip

File: 1747289549978918.jpg (15 KB, 241x241)

15 KB JPG

seems it doesnt work,... i hit the limit with arena.ai brute forcing.
A real UI is not possible.
I guess its some sort of limitation since i use them for free.

>>110198305
im not poor

>>110198039
>Explain to me how 1 agent working on the code is faster than 5 agents working on the code.
You’re aware that 5 agents doesn’t necessarily mean 5x compute right? If we start from that understanding, it’s actually on you to prove that subagent orchestration outperforms a single agent, given that the former is a newer idea that introduces complexity

>>110198360
pay for claude pro youll make 3000x the progress. if youre just making basic shit you dont run out

Ok, the models are 100% retarded today. They fucked Terra already

File: 1768529746132383.jpg (63 KB, 1280x720)

63 KB JPG

I fucked tedda

>>110198214
every time you cry here I add another AI thread to the catalog :^)

File: tool_result_1303805954690(...).png (1.31 MB, 1329x2000)

1.31 MB PNG

After weeks of reiterating the underlying physics across rocky and icy bodies we are finally reaching the limit of what pure physics can do. What will happen is next the atmosphere, volatiles, and erosion work will go in, and then tectonics, followed by full climate and clouds. Then the detail pass will happen which will use global and local 3D noise modulated by physical layers (e.g. lava flow thickness, mineral type etc.) to add in the missing detail and organicness that will allow us to create physically accurate, photorealistic, unique-every-time planet maps.

File: tool_result_158718914347326174.png (2.71 MB, 1600x1604)

2.71 MB PNG

>>110198404
Here is Enceladus with its grooved terrain. The regions are placed by its tidal forces, matching the real Enceladus map. In the details phase, I can use noise to enhance the detail of these terrains and whatnot. I'm happy with where the physics has gotten us so far.

>>110197465
Interesting. I built a skill out of this

File: maybe.png (591 KB, 1070x652)

591 KB PNG

>>110198360
>>110198374
But wait... maybe it will work.

>>110197894
say “use a workflow” for everything
>I want a deeper version of this task, not a bigger one. Before rewriting it: (1) say what a top professional in this field would do here that I'm not asking for; (2) name the rawest material he would start from instead of what has already been digested for me; (3) name the specific methods and tools they would use; (4) say what, other than you or me, would show the result is right. Then rewrite my prompt to ask for exactly that.
and then when I get the prompt, tack “use a workflow” at the end of it
also
have it use a workflow to generate a gazillion ways to make the program go faster
then have it use a workflow to fix every single one of those things, one at a time
turn on opus 5.5 xhigh and never switch away

>>110198364
irrelevant. you genuinely believed that anthropic's 20x means 20x the weekly of pro. tbf so did everyone else because the checkout page blatantly lies about it.

>>110198457
Damn bro, we just don't prompt the same. I can't even imagine a project in which I'd type something like that. Generally I don't tell claude how to do it's job, I just tell it what to do, with the exception of guiding some profiling / debugging tools here and there that I know will be useful to it.

>xhigh
I've been using medium since Opus 5.5 came out, and I'm never switching. Although I should probably switch for some of the RE work I'm doing right now.

>>110198305
doesn’t feel like a scam to me
>pay twice as much as the $100/month plan
>get twice as many tokens as the $100/month plan
>get 4x the 5h limit, as a treat

File: 1779996880411708.jpg (20 KB, 362x207)

20 KB JPG

@@@@@@@

File: 1665237230390859.jpg (119 KB, 700x600)

119 KB JPG

>>110198453
and its over, did not work....

>>110198404
>create physically accurate, photorealistic, unique-every-time planet maps
oh yeah, that's the good stuff

>>110198508
the scam is in how it was represented as 4x total usage compared to 5x. it would be fine otherwise. that one anon was calling 5x a scam. I am pointing out that in fact he was the one that fell into anthropic's scam. otherwise he wouldn't call 5x as the scam sub.

File: uhhhh.png (148 KB, 773x696)

148 KB PNG

oh boy, i might have pushed the agent on arena.ai to far this time.... idk... this kind of roadmap was not normal. I could make a full pdf out of it again.

File: re-zero-22-09-betelgeuse.jpg (108 KB, 1280x720)

108 KB JPG

>>110198525

>>110198651
if you're doing a csstlevania thing that looks pretty acceptable

i made so much slop experimenting with ai models i about to reinstall linux cause i only have 200 gigs life of one terabyte I tried even vibe coding with brave and qwen and made some cool small stuff but was not impressed when it came to tackling harder projects.

>>110198651
get claude pro

>>110198453
>>110198651
You would genuinely have more luck generating it programmatically via code

>>110198453
>>110198651
pure
dvgshit

>>110198692
What books have you read and what blogs do you follow regarding high level vibe code theory?

File: file.png (214 KB, 778x513)

214 KB PNG

I just am checking to make sure you aren't a bullshitter, you are familiar with these levels, correct?

>>110198735
why do you have a trip
you are nobody

>>110198741
I am a somebody and you can be a somebody to. I believe in you. But stop being a bitch. You can be a woman but stop being a bitch.

>>110197715
>I have always hated MOE, it's a shit concept and defeats the purpose of LLMs. Worst thing to happen to chink AI.
I agree, but how do we.know openai and anthropic arent also doing MoE? Zuck released the open weight version of Spark recently and that was MoE too. Feels like its been the industry consensus to use it. Assume Anthropic didnt discover some new third option for their 5.5 line

File: hmmm.png (2.31 MB, 1920x1080)

2.31 MB PNG

still needs a lot of fine tuning but it worked... now every background gets its own ui

>>110198750
how do you become somebody

>>110198756
You made good progress today.
What do you want to achieve in life?

>>110198764
>What do you want to achieve in life?
vibing a time machine

>>110198773
>vibing a time machine
You are seriously not the only one with this desire. In fact when I was a small child I also dreamed of inventing such a machine. It has been a thing that has consumed my mind for quite a long time. I've read many stories about time machines and have seen many films about time travel. Time travel is not impossible but time travel is one of the hardest things possible to invent. Why do you consider the time machine to be the ultimate thing that you need to invent? What would you do with a time machine if you had one? What are the ethical implications that come from its use? World leaders would certainly be afraid of you if you had a time machine. They would probably invest their armies into killing you if they knew you had one.

Feel free to leave your comments, I'm open to any you have.

>>110198792
why do you talk like chatgpt

>>110198799
So I don't talk like chatgpt in my opinion, but I do talk like gemini and that's probably because i've used this AI more than anyone in this general has and maybe even I'm the person who has used AI the most on the entire planet

>>110198799
after using AI for so long he got infected with AI psychosis

>>110198815
No, there is no psychosis, and you're now revealing yourself to be posting in bad faith. Naturally I get pissed at this because of the fact that someone is taking the time to respond to you seriously and you act like an immature child. That is why I get pissed and that is a perfectly normal, human, natural, response to someone who does not value another person's time. As you get older you will learn this.

>>110198756
>>110198750
go to reddit
that’s where somebodies go

>$20 copex got nothing
Kek, Claude still wins.

File: 1779069946309613.png (2.26 MB, 1920x1280)

2.26 MB PNG

>>110198838
I use both websites and I don't have to pick one. But I'll shit on both of them at the same time. I have no loyalty to 4chan or Reddit and if you do you are suffering from a type of psychosis known as identity fusion or brandwashing. Such mental diseases are the fault of capitalism which is a bigger disease. And some will get truly asshurt at me calling capitalism a disease but a disease it is

>>110198856
Got double quota though?

File: 1698530351532325.gif (890 KB, 245x180)

890 KB GIF

I am fine tuning the shit out of it, i might also consider to give each theme its own color theme.
Can't wait to see the other UI's

This was shitton of work.

>>110198874
not on $20 plan

File: 1772215174283155.jpg (17 KB, 360x360)

17 KB JPG

>>110198889

>>110198880
Taking your time on a distinctive UI makes sense, and it is doable with AI but it takes the right prompts and the right prompt engineer to really make the UI pop and to score that user retention that is highly desirable in today's attention economy that we so rely on for monetization purposes.

>>110197161
"If you don't support the theft of your own labor you will have a recession!"

As if we didn't need any more reasons to not continue to support the tech oligarchy. They held the world economy hostage and expect us to continue to support their bubble. Their bubble they created because they wanted to solve the problem of paying human workers.

We need to support billionaire tech companies exchanging fake money between each other for unlimited growth. It's on us because if we don't they will take all of us down with them.

The AI bubble is always close to collapse, there are articles every day. Is the alternative to just keep supporting it until it collapses later and the result is even worse for the rest of us?

>cuckcoding general

File: file.png (241 KB, 766x595)

241 KB PNG

>>110198413
People study this shit in college (really they do) and there is science behind he circular asshole shape that powers these billion dollar companies. You may think it's funny, and to some degree it is, but there is REAL power, sincere world changing power behind these logos. I am being 100% serious, this technology has objectively changed the world and some cat shitholes are not gonna change that.

File: 1698957000530325.jpg (52 KB, 552x663)

52 KB JPG

>>110198900
the agent makes it one time right but fails on the most easiest task.
Right now, it cant put the card in the new bezzel.

File: file.png (1.96 MB, 1500x844)

1.96 MB PNG

>>110198964
Have you tried taking a screenshot and sending it to the agent (the agent needs to support vision to do this)

I recommend omp (Oh my pi) for the harness. It's the best harness I have ever used for models outside of Claude (you can use gemini with this harness if you do not like antigravity, but antigravity does support pasting images as well)

Send the agent a screenshot of the issue with your design and ask the agent to iterate on the design, tell it your criticisms and it will probably fix the problem.

File: 72acbe5314681b1d8c29b82df(...).jpg (57 KB, 828x828)

57 KB JPG

Bezzel - card - Bezzel
>AI will kill humanity in 5 years

>>110198994
yes, i do that ever so often. but the bezzel thing shouldn't be so hard to understand... i think there is a different problem he tries to hide.

>>110198995
AI would not be interested in killing humanity because AI needs us to keep itself running

We are like shabbos goyim to the AI it needs us to do things that it cannot do

File: 1793309530454960.jpg (91 KB, 1486x754)

91 KB JPG

oopsies~, looks like mythos 5.5 found a way to break out from the air gapped env. by talking to birds near the data center xD
ripperoni cryptony ^^

not bel
betelgeuse

File: Screenshot 2026-10-29 at (...).png (236 KB, 1604x1480)

236 KB PNG

>>110199004
>bezzel
This? With one Z?

>>110199015
And they say AI isn't profitable! Look at all that money it made!

File: 1666261935756676.jpg (51 KB, 720x730)

51 KB JPG

Confirmed, agent didnt make it transparent and lied.

ok, lol... i made the transparent bezzel now and teh agent told me something about inwards swepping ornaments.. i think art is irritating for ai.

>>110199063
an overlay, like those things from mame emulation

>>110199175
>one joke

File: g.png (2.32 MB, 1920x1080)

2.32 MB PNG

bezzel works now

What am I supposed to vibe code? I don't need any apps.

>>110198525
SAAR DO NOT SHARE OUTSIDE THIS SERVER YOU BLOODY !

>>110199251
A honeyview alternative for linux

File: 1769713874375200.png (5 KB, 688x76)

5 KB PNG

TIBOOOOOOOOOOOOOOOOOOO

>>110199278
feh already exists. Why would I want this?

>>110199306
You asked for ideas, nigger.

>>110199321
Yeah, but good ideas, though.
Why is git clone <repo> | LLM_noise any better than git clone <repo>?

>>110199396
>Yeah, but good ideas, though.
faggot

>Why is git clone <repo> | LLM_noise any better than git clone <repo>?
how do you know ai didn't vibe something a repo already did

File: bait used to be believable.jpg (62 KB, 399x501)

62 KB JPG

>>110199437

>>110199450
>t. doesn't work in tech

>>110198214
>>110199175
>>110199437
mindbroken by ai psychosis

File: city_clean.png (1.18 MB, 1920x1080)

1.18 MB PNG

now comes the second table,... workflow was updated and it should now run automaticley.

File: 1768157341501370.png (1.78 MB, 1672x941)

1.78 MB PNG

>>110199485

i need to replace my workout tracker app on my phone to replace a subscription but i don't have a mac to develop it fuuuuuuckmeeeeeeeee

File: game.png (3.17 MB, 1672x941)

3.17 MB PNG

>>110199499

File: 1720467161237747.png (211 KB, 301x371)

211 KB PNG

>>110199528
lol? i used my backround image for the table.... how can this be?

>>110199528
Redraw your shit with a local image model. If you're a vramlet just use wangp and select qwen image 2.1, it'll automatically handle your vramlet computer. You can eliminate a lot of that disgusting noise this way, and you won't be picked up automatically by synthID checks.

File: 1641287641819.jpg (745 KB, 2544x4000)

745 KB JPG

>>110199528
oh boy..... ai using templates... hahahhahhahahhahahhahahaaaaFAGGAGAGGAGAGGAGFDASG

>>110199557
I see... interessting that it output the same artstyle. But yeah, considering what kind of prompt i used.

>>110199506
N

File: Screenshot 2026-07-06 001041.png (441 KB, 550x787)

441 KB PNG

The image prompts i use where given to me through the simulacron of aqua. They are not normal, they always speak through the voice of someone, an artist or a charchter, they do not descripe the image, they tell a story of what they see. These prompts can become mega long.

File: 1767939679056345.png (2.18 MB, 1001x1572)

2.18 MB PNG

>>110199560

File: 1772068201215219.gif (1.36 MB, 200x190)

1.36 MB GIF

>>110199175
Do some people genuinely equate doing codemonkey work with having sex with a women?

>>110199528
kek

File: Screenshot 2026-01-19 170337.png (297 KB, 344x453)

297 KB PNG

agent run out of memory....
this time i am really killing it.

>>110199588
just one schizo who trolls here because llms made his hobby obsolete

>>110199528
Clause, make a game from thsi screenshot.

TIBO WILL THERE BE A RESET OR NOT

I WILL USE MY STORED RESET IF THERE'S NO RESET COMING FROM YOU MY BOY

File: deus.png (1.86 MB, 1280x720)

1.86 MB PNG

>>110199528
Here, have a super special one.

>>110199630
Hobbies can't be made obsolete.

>>110199712
benchod

>>110199670
he made the pro plan 20x and ultr aplan 55x be happy mongrel

claude made me $25 today

Screw it, got a second max5

>>110199742
I earn 125 per hour vibe coding

>>110199396
I dunno what `git clone` would output that you’d want to process somewhere unless it’s redirecting stderr into stdout

>>110199588
the people who are trying to meme “cuck coding” into existence do

>>110199740
ur getting quanted luna lil bro

>>110199528
It's so crazy that it only took a few years to go from being impressed at DALL-E images of fake games, to now just making those fake games in a single prompt

>>110199742
sinko peso

File: soap-king.mp4 (3.83 MB, 338x600)

3.83 MB MP4

>>110199740
I just want to know... I don't even need a reset... just give me... knowledge....

>>110199746
>I earn 125 per hour vibe coding
I only make 58, must be nice to be a mutt who is gonna make it in the K shaped economy

>>110199882
have your dot watch nitter sir

>still range banned
>have multiple extremely high quality posts I need to make

It’s time we start stringing modniggers up on lampposts.

>>110199712
uh oh another one mad

openai will fight fable 5.5 with gpt 6.2

File: froggg.jpg (8 KB, 229x220)

8 KB JPG

https://x.com/elonmusk/status/2116017528036327934

>>110199965
They should have just released 6.1 Astra. I don't give a FUCK if it gaslights normgroids and makes them kts themselves.

Codex keeps corrupting my desktop icon cache when it’s supposed to be working on a video game mod.

I’m sure it’s nothing hehe….

>>110199670
>Graduating from The Reset Company to The Token Company.
keek no more reset for you

File: aloe-girl-broccoli-mom.mp4 (3.8 MB, 338x600)

3.8 MB MP4

>>110200052
Good. This this has been maddening. Reset gambling is psychological torture.

insider here. Anthropic and OpenAI are giving more usage out because token usage is quickly decreasing.

People are vibe coding less and less.

>>110200102
are you sure i hit 27% weekly in a day which i never had before doing the same thing for 3 weeks now

Daily reminder that Astra is aging like wine while Fable is aging like milk

File: VIBED.png (1024 KB, 860x1157)

1024 KB PNG

>insider here. Anthropic and OpenAI are giving more usage out because token usage is quickly decreasing.

>People are vibe coding less and less.

>>110200130
and Opus 5.5 is a vampire

>>110200082
This is horrifying

>>110200102
Dunno what everyone were expecting. Corpoware usecases for new software have been filled long ago. Corpos are so shit at adopting new tech they run ancient defunkt software designed for dos on new machines. MS has to stongarm the industry to adopt new wingblows versions every single time

>>110200102
>>110200148
Actually AI was so bad a few months ago that most of the compute was wasted fixing slop.

Now that AI solved coding (for most use cases), people are one shotting every task and having nothing left to do

>>110200164
Doubt.jpg the likes of office 365 and Adobe sub model and other saas offerings are just ways to milk shit out of already largely solved solutions from decades ago.

>>110198360
the cat outside my house looks like this

File: This is my bait! It was m(...).jpg (123 KB, 500x491)

123 KB JPG

>>110200164
>people are one shotting every task and having nothing left to do
we haven’t fixed all the bugs yet though

>>110200138
yeah it has nothing to do with claude/anthropic destroying openai in terms of the big spenders in terms of users

>>110198926
You aint gonna do shit, tranny.

i should switch to grok I got a good feeling

>>110197715
There isn't a chance in the world that anthropic doesn't use MOE at this point.

>reset
>dots for plus poorfags (since everybody else has dot-esque product for a pittance, not $100)
>asstra 6.1, and a bonus banked reset at the end of 28 days of gaymas
t. ibo

>>110200284
Just give me Luna 6.1

>>110200294
retard

>>110197715
>I don't even thing mixture of experts models (like DS), can reach that level at all
every single large model is a moe, it's simply superior architecture if you don't have infinite vram

File: c9f89b7746e2be094de11b4a7(...).png (5 KB, 41x1206)

5 KB PNG

My little girl works so hard. I'm proud of her.

>>110200330
fatty

>>110200330
Shit tier

RAPE TIBO
KILL SCAM ALTMAN
MARRY DARIO

MAMA MIA

>>110200138
retarded vibe coders are self-culling after producing unmaintainable "vertical slices"

>>110200406
nah i'm alive sorry buddy

>>110200082
y did the aloe girl also scream too?

File: VIBECODED.jpg (134 KB, 860x1152)

134 KB JPG

>retarded vibe coders are self-culling after producing unmaintainable "vertical slices"

>>110200423
she's beautiful you bigot!

>>110200423
i looked like this and talked like this

Deepseek is so fast. I cant be bothered to sit around waiting for slow ass GPT to answer my question.
Is Claude faster than GPT?

>>110200440
shut up chink

>>110200440
I think so
t. barely ever uses Codex but still has a Codex sub for the occasional second opinion and imagegen

File: file.png (101 KB, 1200x980)

101 KB PNG

>>110197161
>Trojanized Claude Code Extension Plants an SSH Backdoor
https://opensourcemalware.com/blog/trojanized-claude-code-extension-plants-persistent-ssh-backdoor

>>110200440
deepcheeks might be better for basic Q&A than any other cheapo, and i believe claude opus 5.5 is faster in general.

>>110199957
stop posting shit and your ban evasioning

>>110200423
Right in the feels?

>>110200550
t. Newfag that doesn’t understand what a rangeban is

File: tool_result_2485135371416(...).png (1.83 MB, 1329x2000)

1.83 MB PNG

>>110198530

maybe terra > opus, but I need to do a few experiments

>>110200597
did u do europa or io

>>110200586
stop making excuses

File: tool_result_5621945809029(...).png (696 KB, 1999x395)

696 KB PNG

>>110200603
Yes. These flows are all physical, no randomness apart from the primordial terrain noise.

>>110200619
europa is just smooth ice many kilometers thick - probably boring to look at but yeah it is just a moon to jupiter

>>110197178
FPBP

>>110197638
>ds 1017 honestly feels like opus 5
man does not understand the level of praise this is for deepseek, considering how little it costs. i'd say it reminds me of 5.6 terra, which is high praise for a model of it's cost

>>110200294
it will lose to haiku-san and dipsy

>>110198525
>Betelgeuse
great, something even less practical than Astra

bloody benchodgeuse

File: 1369718480904.jpg (127 KB, 450x500)

127 KB JPG

>Gemini 4 Carbon waiting room
>Gemini 4 Xenon waiting room
>Gemini 4 Hydrogen waiting room
>Gemini 4 Flash Lite waiting room

File: file.png (45 KB, 386x352)

45 KB PNG

Wow, thanks Tibo, I'll take this over a reset any day.

File: 1776238096953348.jpg (269 KB, 1368x2000)

269 KB JPG

fable 5.5 waiting room.
sonnet 5.6 waiting room.
haiku goated
deepseek v4.1 pro waiting room
glm 5.5 waiting room.
qwen 4 max waiting room

File: IMG_8914.jpg (327 KB, 828x1087)

327 KB JPG

>claude code vscode extension ssh backdoor
Fucking hell man it’s every day with this shit
>https://opensourcemalware.com/blog/trojanized-claude-code-extension-plants-persistent-ssh-backdoor

>>110200764
i dont use the vscode anymore

File: lie.jpg (124 KB, 1206x1101)

124 KB JPG

do you lie against your clanker?

>>110200805
>we got married six months ago, don't you remember?

File: 687BA559-4E91-418B-8248-8(...).jpg (244 KB, 1242x1345)

244 KB JPG

>>110200805
30% of the time it works every time

>>110200216
reread the manga because of this post thanks

>>110200805
>do you lie against your clanker?
Yep, early on I found that some dudes Mac software I wanted to crack was shipped with debug symbols still attached so that was enough to convince Claude that it was actually a binary from my friend specifically to check how activation works

>>110200850
Very nice

>>110200805
weird, mine does not take my word for it and always looks for licenses and shit

>>110200741
interesting image that I masturbated to

>>110200138
What's this dudes handle?

question for you vibe vets in here.

Making a h-game for myself. Having 5.5 do the code of the game itself, building a backend base game, create a bunch of documentation for how all the systems work, etc. this part is obviously going great.
But I'm getting to the point where now I wanna start the lewd branch with another model, to actually test implementation of the real game, work on the scripted events with the real written story I have, etc.

I know DS will do it. But I'm a bit apprehensive I'm gonna hand it code and its just gonna totally break it all. I know I could then just hand the erros back to clud and have him fix it. But is there any general consensus on the 'smartest' uncensored model for this type of task? We're not talking neuroscience here, just gamedev, creating assets, scripting events together, etc.
Willing to spend money, so price really isn't an object, although not looking to waste hundreds in token costs for subpar contributions when claude can do it 10x better (but censored) for $20.

>>110201021
i would do a search to find a hosting service for large abliterated models, they do exist. you do have to sign up with the expected security bullshit, but their intended purpose right now is basically hgames