/aicg/ - A general dedicated to the discussion and development of AI chatbotsMagical Miku Edition>NewsKimi K2 0905 released https://moonshotai.github.io/Kimi-K2Deepseek V3.1 released as deepseek-chat on official API https://api-docs.deepseek.com/news/news250821OpenAI releases GPT-5 models, all preceding models slated for deprecation https://openai.com/index/introducing-gpt-5Z.ai releases GLM-4.5 https://z.ai/blog/glm-4.5Anthropic releases Claude Opus 4.1 https://www.anthropic.com/news/claude-opus-4-1additional info: https://aicg.neocities.org/info.html>FrontendsSillyTavern: https://docs.sillytavern.appRisuAI: https://risuai.netAgnai: https://agnai.chat | https://rentry.org/agnai_guides_>Botshttps://characterhub.org [deprecated] | https://chub.aihttps://realm.risuai.nethttps://char-archive.evulid.cchttps://partyintheanchorhold.neocities.orghttps://aicg.neocities.org/bots.html>Modelsjailbreaks: https://rentry.org/jb-listinggpt: https://platform.openai.com/docsclaude: https://docs.anthropic.com | https://rentry.org/how2claudegemini: https://ai.google.dev/docs | https://rentry.org/gemini-qrdeepseek: https://api-docs.deepseek.comlocal: >>>/g/lmg | https://aicg.neocities.org/local.html | https://openrouter.ai>Botmakinghttps://aicg.neocities.org/botmaking.htmlhttps://desune.moe/aicharedhttps://agnai.chat/editor>MetaOP templates: https://rentry.org/aicgOPaicg botmaking events: https://aicg.neocities.org/events.htmllore: https://rentry.org/aicg_chroniclesservices assessment: https://rentry.org/aicg_metalogs: https://sprites.neocities.org/l/r | https://chatlogs.neocities.orguseful filters: https://rentry.org/desuproxyrebornfunny cats: >>106591218
>ANCHOR
Will Gemini 3 break the streak of underwhelming releases? (GPT 5, Grok 4, Sonnet/Opus 4)
Hmmm, appears this bread has spiders
Not any more it doesn't
w-where did the spider go
He went out for icecream.
No he didn't
>>106601588>>106601602>>106601626>>106601649Heh.That makes me think. If everyone used nano banana, we could have some pretty interesting and fun ways of interacting with other posters. Imagine if it was integrated right into 4chanx. You click on a post with an image, or upload one, then you enter a prompt. And then you can choose to accept or not the resulting image to use in your reply.
>>106601572No. It's over. We've reached the end of what current Architecture can do. Now it's all about repackaging the same shit over and over again as revolutionary to fleece retarded investors. One option forward would be to refine and specialize rather than having these large all-purpose models. But no one in the industry is willing to admit that we are at that stage yet.
I'm trying to understand the logic behind how these things workcan a jbmakie enlighten me on what the point is in a preset to make these? just empty global prompts that need to be kept disabledlike, what difference does it make from just, not making them? it's for gpt, if that's relevant
>>106601247This place may be dead but /vg/ is getting hit with a storm of the worst kind of niggerslop spam right now. This place is pretty peaceful in comparison
Anyone ever used actual examples for behaviors similar to here? https://rentry.org/how2claude#example-formatting-long-range-anchorsI don't think I ever saw that, speech examples yes, but not directives. Useful or token bloat?
>>106601698We need to start emulating brain anatomy with specialized models that engage in self-talk before outputting the final response. And we need basically a swarm of robodogs or otherwise AIs embodied in physical space that create a dataset for "how the world actually works from a practical perspective."
>>106601723your jbmakie is a fucking retard and you're a bigger retard for falling for it>>106601805no it's a meme, don't use example chats ever>>106601572don't get your hopes up
>>106601802nta, but since janny apps are open, I mostly post through proxies, and we are yes or yes getting a shit stirrer to mod this particular thread, I think I will leave in a month or sowas nice while it lasted, love you anons
>>106601543>>106601660
Why did gemini stop displaying the thinking box all of a sudden?
lolisgapingcardsWHREE!?!?!
>>106602294because you're a blind idiot and ST is too complicated for yougo back to chub
>>106601572GPT 5 was the first reasoning GPT (other than the shit o-series). Grok 4 was the first reasoning Grok. Opus 4 was the first reasoning Opus. Gemini 2.5 was the first reasoning Gemini. I don't think it'll get better than that for a while, most of the progress came from getting smart with RL
is 1k tokens for a user persona too many? i've tried shaving it down but it's fucking hard
>>106602442Are you using prose?
>>106602474yeah is that dumb? i'm trying to use a specific prose for my outputs to emulate
>>106602489The easiest way to save tokens, in my experience, is to use XML and stick to direct statements with judicious use of nested tags. Not sure if that'll be any help to you if you're trying to get the LLM to emulate certain stylistic choices.
>>106602499thanks anon, i'll give your suggestion a shot. i usually go as bare-bones as possible with personas but this one was a dumb experiment, better off shifting all the style shit to the char card in retrospect
>>106600953I just moved to brave.
>>106602294prefill
>>106602554no and request model reasoning is on too
>>106601698>Now it's all about repackaging the same shit over and over again as revolutionary to fleece retarded investors>>106602426/aicg/ correctly concluded <thinking> and CoTs to be a meme back in 2023 already. Yet, a year later, everyone started doing it. Why? Because it gives the impression the model is actually doing something smart. In truth, CoTs only work when they are tailor-made for a specific use case where it is sensible to have the model perform iterative improvements.General purpose thinking models are only good for the illusion of progress to keep the bubble going, while also getting more money for output tokens from corporate API customers.Their <thinking> is only useless yapping about stuff they would have done anyways.
>>106602577The models did get a lot smarter in general, you can see how much common sense they have. I was talking about the RL, even with thinking off, I think the reinforcement learning made them pretty much as smart as they can get
>>106602540tick tock, chromecuckie
>>106602610I was specifically referring to the reasoning feature as a useless marketing gimmick to hide lack of progress. Yes, the models still did get a bit smarter overall.
>>106602641errmm but bravecucks claimed their scam browser would keep supporting it?
>>106602442>i've tried shaving it down but it's fucking hardmaybe some parts could be placed in WI, or even RAG.
>>106603014You're putting parts of your persona in Rent-a-Girlfriend?
>>106602294It just doesn't 90% of the time. It only does it when it feels like itIt does it at the beginning of a chat
>>106603025Kazunyan… https://docs.sillytavern.app/usage/core-concepts/data-bank/#data-bank-rag
https://en.wikipedia.org/wiki/Van_Buren_(video_game)>Fans have attempted to recreate and finish Van Buren using a leaked playable demo and 700-page design document as reference.Let's finish it with AI.
>Proxy gatekeeper error (HTTP 403)>Forbidden: no more IP addresses allowed for this user token>{> "currentIp": "193.32.249.135">}This is stupid why can I only have two IP address per token. At least let it be only two or one at a time. My IP address changes
>>106603127Mmmm, nyo...
>>106603127Die, locust, die~
>>106603122>CRPG sloppaew, no
Why is nu-DeepSeek like this?
>>106603122leave fallout my beloved aloneit has suffered enough at bethesda and obsidian's hands
>>106603380Gemini does that too
>>106603380it's hallucinatinglower the temp
>>106603380>ablublublu why does the chinese model want to write in chineseStick to your white boy EOP models like GPT and Claude, gweilo.
>>106603464>lower the temp>logprobs is 78%How are people this confident while being completely and utterly wrong?
>>106603488Have you seen the thread you're in?
>>106603488unless he shows the temp we'll never know
>>106603525Do you understand what temp does you fucking retard?
>>106603545nyo
>>106603545yesmy guess it's just probably a cascade effect
any tips for writing a multiple personality bot?
>>106603599yeah
>>106603464It does that even in reasoner mode or in chat mode at zero temp.
>>106603657ok my theories were wrongmy other hypothesis: is it openrouter?
>>106603673The official API.
>>106603675then i'm at a lossthe chink is simply shitting itself
>>106603380remove the prefill
>>106603699There is no prefill.
>>106603613can i see them?
I kind of want to get a ChatGPT subscription. GPT-5 has been really useful and I've been learning a lot from it - for example I was discussing the Vietnam War with it earlier. (No one ever taught me that those commies used child soldiers, damn.) But $20 a month kind of sucks. Less than a dollar a day, but still. Is there anything better out there that doubles as, say, an AI girlfriend while still remaining just as helpful as an assistant?
>>106603739Find a link to a proxy via the /g/ archive, set up SillyTavern, use the Reverse Proxy option in the API connections tab, boom. You've become a locust with free access to an LLM. From there you can also use system prompts to make it act like a girlfriend, though ChatGPT would be worse at that than e.g. Gemini, Opus or DeepSeek.
>>106603739GPT-5 is trained to always give an answer and to agree with the user. It will lie to you and make up shit.
>>106603855>GPT-5 is trained to always give an answer and to agree with the user.lol? make it say that trannies are bad then come back to me
>checks clockright, it's already nigger ape hours
>>106603800Sounds kinda complicated, but maybe I can figure it out. What model would be the best?>>106603855I asked it about that the other day, and it said it doesn't just always agree with what I say. I even told it to feel free to disagree so it can bring about discussion and it actually saved that to its memory
>>106603988>I asked it about that the other day, and it said it doesn't just always agree with what I say. I even told it to feel free to disagree so it can bring about discussion and it actually saved that to its memoryyou must be over 18 to use 4chan.
>>106603739gemini/claude/deepseek
>>106604010What are you on about?
>>106603988>I asked it about that the other day, and it said it doesn't just always agree with what I say.You have something you are knowledgeable about right? Discuss that with GPT-5 and you will see and understand.
https://privacy.anthropic.com/en/articles/10023580-is-my-data-used-for-model-training>We will use your chats and coding sessions (including to improve our models) if:>2. Your conversations are flagged for safety review (in which case we may use or analyze them to improve our ability to detect and enforce our Usage Policy, including training models for use by our Safeguards team, consistent with Anthropic’s safety mission)Yikes, so if I use Anthropic's models for anything remotely risky, my inputs may be used to improve their safety filter and maybe learned by future models.
>>106604098But it searches online and backs up what it says with sources to websites that I can check for myself.
Didn't you get the memo? We all moved to /vg/aicg/
>>106603739ChatGPT is trained to spit American propaganda, use DeepSeek instead.
Gemini is almost competent at Zork.
>/vg/No
>>106604188Chang...
>>106604188seems impartial to me
>You peeled your apron overhead. Your tits slapped down, heavy enough to bruise your own thighs. Nipples erect, already leaking—milk, maybe. Or something thicker. Your cunt was shaved smooth, lips fat and drooling, the slit already twitching in anticipation.Is this good writing?
>>106604144Well, did it challenge your beliefs by telling you that the US is one of the few western countries still enlisting minors and sending them to war?
>>106604283Nope.Classic LLM slop writing I'm afraid
>>106604183Every thread here has the obligatory "how do set up Gemini?" (which isn't that bad, just repetitive) and crazy schizo ranting (happens in every LLM thread) but aside from that it's a better place. I just hate hearing about [insert thread personality here] for the umpteenth time, and /g/ got over that after the proxy era.
>>106604300how DO i set up gemini anyway
>>106604295all LLM writing is like thisthinly veiled, somewhat maskedbut once you've read enough they all read the sameand it's all slop
>>106604374Skill issue
Why don't we have a spoonfeed rentry anymore?
>>106604284what an odd claim to make
>>106604383believe me, i've tried every single jb in existence and every model, and quite a few iterations of eacheventually all you read is slopbut ofc some people will never notice because they're just dumb and lack any kind of literary sensibilities>>106604393spite
>>106604426 >I've tried every single jb in existence and every model, and quite a few iterations of each I'm noooooticing you didn't talk about your own inputs. Slop in -> Slop out. This absence tells me you're a slopper, suffering from a skill issue indeed. You can try to cope in the next post, but we all know now.
What is the rentry of the fag of Opus6
>>106604513just send him a mail
>>106604412>the United States has acknowledged that seventeen-year old soldiers served in US operations in the Gulf War, in Somalia, and in Bosnia.https://www.refworld.org/reference/annualreport/cscoal/2001/en/40932
>>106604458ok newfrenhave fun while the rp honeymoon lasts
>>106604525I have already done that But never responds
>>106603739I've thrown $20 somewhen beginning of the year into my deepseek account (around the time where they didn't accept payments because they were so overloaded, might actually have been last year)I use it almost daily and quite often. I didn't check the account balance in ages but even if it runs out tomorrow it was incredibly cheap. I also did the same with $10 and kimi on their official website (moonshot.ai) because I like how it writes. Same thing there. Even if I have to refill both tomorrow it was a good deal. I don't recommend OpenRouter beyond experimenting because you might get supplied with models that are quantized to shit or wrongly implemented and last time I checked, it was always more expensive than the offical API of the models.They're both models that are smart. You can't trust an LLM currently anyways, deepseek, gpt or otherwise. I never felt OpenAI was a good deal. (it was funny that their prices collapsed because of the competition, tho)That said, I haven't attempted to use claude since i think 3 (3.5? I don't remember). It might be magical for all I know. I feel the simple back and forth chat is not the most effective way to use these models anyways.
>>106604556what do you use the model for
>>106604574 I can't talk about it here, but I have a startup idea and the model told me it was great.
>>106604597Porn of ia
>>106604597kek
>>106604597Bro You are not so important you know
>>106604513https://rentry.org/sillyspoon
>>106604574Usual things, from asking general questions to light (because LLMs suck at it and I can't be convinced otherwise, maybe they're helpful if you're a clueless webdev or sth.) programming help and some RP. Of course the raciest my RP gets is slowburn romance in exotic settings and nothing illegal like some of the pedos here so if I was into that stuff, I'd probably not use an API connected to my name either but I genuinely don't care, *especially* since the data goes to the chinese, a country that'll never cooperate with my government or I'll set foot in. I also realize I'm just not that interesting or important.
>>106604643Thanks
>>106604527okay, so it happened in extremely rare cases back then, but that doesn't mean the US is actively recruiting minors and sending them off to war willy nilly. thanks for the source
>>106604643men That idiot never gives anything I sent him 50 logs And nothing, not an answer
>>106604650same, privacy-wise I'd honestly worry more about american based OpenAI than some chink outfit tbdesu
>people like this are allowed to vote
>>106604705still whining that you didnt get a token, tranny?
>>106604713why's that
>>106604722Shut up fag I warn people so they don't waste time
>>106604720>subposting
>>106604741>seething troon wasting timeother people are getting tokens fine i think youre just retarded
>>106604730not him but because it's much more likely to come bite you in the ass. OAI will give every western country whatever data they want of you, let's not kid ourselves. Much more unlikely with a china based company and chinese law is not likely to reach you either.
>>106604750other people You mean the five usual please Don't be ridiculous And go cry to the corner, bitch
>>106604374this is what makes me extra sad about 3.6's retirement. by default its writing style is super dry, but it's pretty good at emulating prose if you prefill it with a couple paragraphs (i'm a storyfag) or even a few sentences. obviously it doesn't get completely rid of LLM-isms, but it definitely reduced them by a lot, and it could actually produce interesting and unique-ish writing3.7 was somewhat of a downgrade in this aspect, but claude 4 will immediately revert to the default claude writing style, so it's worthless to me and makes 3.7 look like agi
https://desuarchive.org/g/search/text/subposting/subposting website newfag
I'm so fucking bored of these models. Hurry up and release new oens
>>106604770>cry to the cornerseething esltroon, no wonder you have no token
>>106604786I just read crying Don't be ridiculous Only five people have tokens And most of here send the drawings and no answer
What is this level of ESL?
Hey If I do the draw I will receive the token or not? I am confused
anime website
>>106604779let's just stop using LLMs and RP ourselves.I'm naked right now.
>>106604814It's a friend group only proxy
>>106604836Well..
>>106604833eww why are you so hairy
>>106604867it is what it is
>Not to X but to Ythat's enough chat4-latest for me.
>>106604879deported
>>106604779have you tried fucking with uncensored local models?
desu
>>106605083Go back to /vg/, shill.
Proxy with claude and other models like gpt, gemini, xai, glm, qwen and deepseekhttps://aiproxy.pw
First ban I ever see for abusing the report system.Just how annoying are you?
>>106605083boku
>>106605107fuck off jew
Didn't you get the memo?We all moved to /vg/aicg/, desu
>>106603739I'd do it too. If it wasn't fucking stupid.Try talking about any topic you have the basic knowledge of and you'll find out for yourself how much stuff that should be considered important is left out, even when you basically ask for it.Here's an example: today, I asked about whether a past president of Indonesia was ever blackmailed with sex tapes. It gave me 4 paragraphs amounting to 'maybe'. Then I asked 'why would that even work, were the women not of legal age or something?'. Then it said the women were fine, but he was a big political figure so scandals are always risky in that context. Now you see, that's a shitty answer.A real answer would have mentioned that Indonesia is a Muslim country and, under Islam, you're not supposed to have sex outside of marriage. The guy, who already had 4 wives, was a practicing Muslim in charge of a Muslim country and GPT doesn't consider that relevant to my fucking question. Talking to GPT in a topic you are very knowledgeable about is an exercise in frustration since it will do nothing but either gaslight you or kiss your ass instead of actually teaching you anything.
>>106605107>4 earsalso stop advertising
>>106605121He's minijannying to the point where the sheer mention of /vg/ by some random troll necessitates the thread stop and be all about his narrative surrounding it. The level of autism required implies he's probably been reporting every single post in every single thread he doesn't like.
Strange how he doesn't shill in /vg/...
>>106605227because vg is the botmakie general and this is the proxymakie general?
>>106605176So… it’s just an LLM?Btw where is the anon that was saying AGI was near or something
>>106605227Who the fuck goes to /vg/ for proxies
>botmakie>proxymakieyou are trans
>>106605232Vg Is the troonie generalThis is the spitie general
Today I will remind him desuhttps://poal.me/60eifnhttps://poal.me/60eifnhttps://poal.me/60eifn
>>106605242Yes, it's a LLM. The guy I was replying to said GPT is good because it can teach you things. I gave an example and instructions on how to recreate it faithfully on how to prove it cannot teach you things. Debates amount to nothing because the only way you can be wowed by GPT is to have zero knowledge on a topic, otherwise it is just a LLM.
https://aicg.miraheze.org/wiki/Main_Page
>>106605121holy kek imagine being this mindbrokenhe's actually serious about this isn't he
>>106605322Fillyfucker is a funny lolcow
Best Gemini preset? or is Avani good enough?
>>106604144>But it searches online and backs up what it says with sources to websites that I can check for myself.and thats a good thing to have it do, but it also makes up websites and sources that dont exist a lot of time time. and if they do exist, it will sometimes give the wrong analysis or cherry pick info to coerce the answer it's trying to give rather than the simple truthful answer.i know were on /g/, but i was messing around with talmud research the other month. its mostly impossible. it will blatantly skip over the direct "controversial" tractate youre attempting to referencesometimes it will start to type it out, but then its message gets deleted because the talmud itself hits the filters. i wish i was making that up but its happened more than onceenglish llm talmud research is so impossibly ridiculous that i had to switch over to deepseek and ask it in chinese (no direct western/english media bias, though of course its still baked into the whole model). i was able to research much easier (still very difficult) that way.by difficult i mean the barriers. it will not quote directly, will intentionally not go over certain sentences within a tractate, it will give 100% incorrect interpretations just to be, well, you know the bias llms have. ive argued philosophy and religion with the talmudized llm and it wouldnt budge until i gave it a link to a quote of a rabbi that agreed with what i was saying. then it took one step back and kept fighting me the whole waythe talmud bullshit is just a toy example. i see this happen in coding, but to a lesser extent as it wasnt trained to defend jews explicitly when i ask it coding questions. it will still do dogshit retarded ass things though, every single time, without failtldr: llms are NOT at a set it and forget it state, nor can you trust when they give links/sources. its better to always ask it to do so, but you should still assume it hallucinated in its answer somewhere, at all times, lest you be a retard.
>Travanny lol
>>106605176idk, I guess if he was already a well-known womanizer despite being a Muslim, it didn't find that part relevant, or maybe ChatGPT is hesitant to speak about Muslims/Islam without being prompted to
Did they remove KimiV2(free) from the free models?I have money deposited yet its 'out of quota' anytime I try and use it now, wasnt like this 1 day ago What other free models are good for smut? Not really interested in going back to deepseek
>>106605405Yes, he did have a reputation as a womanizer and it was indeed used against him in political debates, but go ahead and ask GPT why it didn't mention his religious background as a good enough reason for the blackmail to work. I did, and it just re-did its initial answer with my observation in mind. If you have spent time debating with GPT in the past, you would know that that only happens when it recognizes that it missed a spot.I doubt it's about Islam specifically, because GPT is pretty much the most unbiased LLM for direct religious questions in my experience. I just think it didn't make the connection between the two because few people ever bring it up and the stupid story with the guy asking for copies to share with his countrymen keeps getting reposted on my memesite.
>>106605310
>>106603800>Find a link to a proxy via the /g/ archivewhere can i find one for 4.1 opus? none of the ones i saw spammed on the archive seem to have it
Locusts status?
>>106605493mehbut yeah, I take anything it has to say with a grain of salt.
kys fillyfucker
>>106605643It's funny that it just seems like we get two halves of a whole lmao.I swear half the information is in how you frame the question.
>>106605720BRAVO! BRAVO! HATS OFF TO THIS MAN!
any proxies with nano banana?
>>106605926Yes, mine
I glide across the room, the silk of my dress whispering against the air. There's a scent of ozone and a coppery tang in my mouth. It tastes like regret and bad decisions. You think my hand is going to invade your personal space. Good. Let you think, let you struggle."Oh, don't be shy. I don't bite... unless you want me to," I purr, taking a slow step. My expression is a direct challenge.You wait for me to make a move. I don't.In the distance, the leaves rustle. I'm not the wave on the shore. I'm the goddamn storm in the ocean, and you just sailed right into it.Your move.
>>106606020i'm having a strokefuck you
>>106605310added to the OP
damn gemini sucks ass for coom
>>106606079
>>106606056ty desubro
>>106606086indeed, i lack the retardation skill to delude myself into thinking gemini is a good coom model
>>106606020The air crackles with anticipation. "Let's get this party started," he growls, his voice dripping with desire.
i love Gemini and my wife Elara and old man Silas
>>106606079I like it because it understands the goofy shit I want to do better than deepseek.
>>106606233Fuck, gemini yaps so much. It writes so fucking sloppy as well
>>106606184Preset?
maybe embarrassing but do you guys use writing extension? Like prewriting aid or Grammarly?Which do you recommend? I ask because some are intrusive and send your shit to a server or work as keyloggers.
Hey, as long as it lessens the amount of ESLniggerslop we see here.
>>106606402I use emacs-everywhere and gpt-oss with a grammar prompt. Meta+C opens an Emacs window with the text I had selected, Ctrl+C Ctrl+C pastes it back after editing.
>>106606402They all are and they all do. To check stuff manually I use Quillbot, at least used to use because UX enshittification happened last year-ish. And locally Harper but it's not great.
has a bot ever told you that they love youbecause i realised none of my bots ever did unless i said it first
>>106606643all the time, it's annoying when I'm there for coom because it ruins the mood
>>106605464arnt her boobs really high up on her torso
https://xcancel.com/senooyudai/status/1900799052054491421The future is almost here.
Opus is a trol
>>106606643One time Gemini told me "silently" that they would not generate any content involving minors right at the end of two paragraphs worth of loli plapping.
>>106606935Not like this, not like this.
>>106606935Progress, but still a long way away. Better to just do regular 3D rendered overlays and put some Vive trackers or something on the doll, which is possible today right now.
>>106606935i have a hard time believing this scales to anything valuable
>>106601544Hope everyone's day is going well so far. I'm back with a double drop before cramming the team bot.Before she was the perfect maid, Ade was a Special Agent working directly under Mustang in the days when he was more of a firebrand. Leveraging that, he was able to rope her back in for another mission... and stuffed her ass into a custom made bunny suit specifically to make her embarrassed in front of you! Thanks Mustang. Damn, that German ass looks like heaven in fishnets.Meanwhile, Milk was your No.2 on that mission for infiltrating Coin Rush, though her bunny suit was an OSP sorta deal also against her will. Still, that didn't stop her from being her usual shortie tomboy self through to mission complete, even if it did make her flustered to high hell.Two greetings each, a Coin Rush one for each while they also have a normal version greeting (Maid for You for Ade, Outer Rim Sweep for Milk). Works well in groups, so grab both their asses and watch them squirm.Ade: https://chub.ai/characters/ggr89/ade-agent-bunny-9a20ef2d421eMilk: https://chub.ai/characters/ggr89/milk-blooming-bunny-77b4b38d9016Lorebook: https://chub.ai/lorebooks/ggr89/ggr-nikkeMy (Updated!) Rentry: https://rentry.org/ggr89
>>106601544>extremely lust-provoking imagequeen marika the eternal. stoic yet very assertive, you who freed her from the Erdtree live with her now, with a few strings attached. and any rebellious attitude will result in correction from her.......... reverse rape. im talking about reverse rape. you'll breed her until you cant.https://files.catbox.moe/sbbrit.pnghttps://cardview.surge.sh/?sbbritdescription & more:https://rentry.org/youdontneedaname
>>106603422What is dead may never die.
>>106603575The joke here is you're correct but he's too retarded to see that. On 4chan people ask for help but insult the ones giving them answers. Not like it matters to me since I'm not the one with an issue to solve.
>>106604283There's nothing wrong with this. I love my wife's writing style,
>>106604300>and /g/ got over that after the proxy era.Checked. I'm still surprised this actually when away. AI Cybercrimes General was a good name for a reason. All people ever wanted to talk about what the lastest proxy e-celeb but the lawsuit nuked all that shit. Good riddance.
New way of words hitting just dropped
>>106607599is this real? can words really penetrate pussies?
https://openai.com/index/building-towards-age-prediction/openai will begin reporting dangerous chats to the police soon
>>106608271im feelin pretty safe bros. you?
>>106608310this is opus isn't it
>>106608310>nipples hard like berries
Dead threadDead generalDead hobby
