Which one do you use?
usecase for grok?
qwen localno need for anything else
>>109104534none of them. kys, normalfaggot
Grok and sometimes Qwen
Le chaton fat (mistral ai)
Fuck AI
>>109104534Gemini 3.1 ProBest storyteller IMO
>>109104553Being up to date rather than some knowledge cutoff from 2024 and half-assed websearch added to it
Chatgpt because I'm not a programmer and it seems better.
none of them because IM BROKE, BITCH
>>109104953You will be able to at some point
>>109104534I use whatever that crap is when you do a google search. It sometimes has the answer I'm looking for.
None of them. I mostly use “Bussy”
>>109104534koboldcpp
>>109104534Kobold AL (anti-luddite)
Claude for code (worth every cent)Nano Banana for images
im paying for claude unironically
ChatGPT for general queries Claude for code and studying
>>109104553Memes and shit posting.Use case for asshole shaped Claude?
>>109104534Claude and Gemini
>>109104534Nobody uses copilot. I've heard people use grok for AI slop port, no idea how they get off to that. Claude app on phone is pointless unless you use dispatch or trying to make it work with your photos or screen shots. Same with geepeety and gemini and deepseek apps.I mean, take a photo of some small font verbose bullshit on a random label and make it OCR it. That's the use case. Goggle lense can do it too and it's built in the AndroidOS, afaik, or preinstalled. So idk, no use case for your pic then.
>>109104534Gemini Pro with Thinking is good for programming but it's very limited in terms of free usage. The Gemini flash limited is pretty bad, it screws up/removes comments, adds extra newlines everywhere, and forgets functions when modifying the code. My work provides unlimited access to GPT 5.5 Thinking via Copilot so I just use that for programming and it works great. I use (free) Claude Sonnet Medium with Thinking for automatically analyzing numerous PDFs from a specific website, it's extremely competent and probably the best AI out right now.
who gives a shit they're all the same
>>109104534I get copilot for work, but privately I only run local inference.Qwen3.5-a3b runs pretty good on 12GB Vram + 32 GB Ram
I use grok to generate smut. Is there something better?
>>109105673having sex
>>109105691anon I'm an 5'5" disabled ethnic dude
>>109105698oooof. rip in piece dude
>>109104534started using gemma4 locally and deleted any accounts i had on these big platforms. good enough for me
>>109104574>>109105737>>109105745Based local LLM users. Qwen and Gemma are rock solid.
>>109104534Mostly this right now. I did like grok too, but then they pulled the thinking mode from free users. It was really good at doing web searches and reading websites, which is pretty ironic since Google has the biggest search engine, but Gemini is really lazy trying to use it.
Bard. But only as a search engine through browser.
>>109104574local qwen is fine until you need to do something that isn't a toy project.
>>109104534Mistral because I want to support alternatives to the American LLMs.
I use Claude and ChatGPT. Why is ChatGPT so much faster though? I feel like Claude gives better answers, sometimes, but it's also waayy slower.
>>109105829based
On my phone? None of them.>>109105140They are trying to take away fucking phone jailbreaks and sideloading, and you think you're gonna get an AI sex android? lol, lmao even.
>>109104534Grok for retarded shit and really hard searches.google turned into shit so you need an AI thing.
>>109105852amazing that google is unusable now. like not even hyperbolic contrarian unusable, like actually worthless for a lot of searches.
>>109104930> using grok for that? sentx.ai actually handles it better without the fluff.
>>109105856I use brave for searches now.
>>109105856like, totally, like
>>109104534Gemini is my main one ever since 2.5, it's really good and consistentWhen I really want an answer, I paste the prompt to Gemini, Claude and GPT to compare all the answersDeepSeek was ahead for like 2 weeks and I haven't used it since thenGrok was good for uncensored stuff once but now it's super easy to jailbreak an AI so it's not necessary and Grok got super nerfed and censored after xitter bikini drama anyway, it's useless (new video models were especially awful compared to older ones)Copilot I haven't really used since they rebranded Bing as Copilot, Microsoft really fumbled the ball like they did with Internet Explorer and Skype, it's the worst AI despite Bing/Sydney being so ahead of the time at some point. It has the ugliest, sloppiest UI I've ever seen, model feels super retarded compared to all the other chatbots. Also Microsoft forced it everywhere which made me want to use it less, and image generation hasn't been good since first week of DALL-E 3 back in the day.I also use>Sakana AIthis is a japanese chatbot, I use it for searching japanese sources, very cool but you need a japanese VPN>Notebook LMthis was legit a game changer for me when studying for exams, turns every pdf into podcast or video essay that you can listen to over and over at 2x speed to passively learn stuff.I don't use>YandexGPT / Alice AII was so hoping that russians would make a great uncensored model, hell I remember Alice AI in Yandex Browser being a thing before ChatGPT even existed but they fumbled massively. I think Putler send all talented programmers to get their balls cut off in the war or something, it's so bad and extremely censored. Most disappointing AI ever>PerplexityThey're giving free premium everywhere but it never really clicked for me, other chatbots are as good in searching stuff. It's kinda like Cursor in that it was first to do this specific thing but it's no longer necessary.
>>109105974Oh and I also use Google Antigravity (the normal app and IDE, I like the UI it's very pretty) and Reddit Answers which is like AI that searches reddit which is pretty underrated and useful because normal reddit search is total trash
>>109105610I use Copilot, Claude, and sometimes Grok. I never pay. Free tiers works well enough and Microsoft are sinking billions into making it available. Uncensored local models are fun to play around with too.
>>109104534none of them
>>109104534GLM 5.2It's similar to Opus in performance for my usecase and ten times cheaper. I'll lay off Claude until (or if ever) Fable comes back.
Grok and ChatGPT are somehow mogging Gemini and Perplexity as a search engine even though it should be the other way round
I cannot use grok any more but deepseek is pretty cool for general stuff, ChatGpt does some stuff to a really advanced level but the devs have fucked it in other pays. Perplexity is actually really good for drilling into details on some subjects. Never tried co pilot as Microsoft just have the monopoly on windows and other than some nerdy improvements very rarely deploy new best in class products, bing is still terrible. Gemini is built for ad friendly contect. Never tried claude.
>>109106229I must add, I only really use it these days for drilling down into topic details so I can write report faster. I have used to to make some schedules but how good that is seems to change according to whatever scandal has just hit. I'm not coding or anything with it.
>>109104534I use the chatgpt web interface but it's really dumb for any real questions so i only use it for code api questions like "how do i make a utc date".
>>109105745>Qwen 3.6 27B for coding. All you need. 100% localYou wish (and I wish, too). Tell Qwen 3.6 27B to implement a new feature on an enterprise monorepo and see it destroy it in minutes.So far Claude, GPT 5.5 and Gemini 3.1 Pro (barely) are the only ones that will one-shot everything I throw at them without babysitting them as long as they have the information they need.Local Qwen will fumble tool use and will get stuck in reasoning loops. At this rate, it looks like local with just one GPU will never be usable seriously.
claude and chanLGBT
>>109105974can you share some ways to jailbreak gpt 5.5? forums? discords groups? darkweb?
>>109106919don't know about gpt but gemini models are like completely uncensored in antigravity if you just use .md rules (and I think you don't even need them)
>>109104534Lumo and duck.ai
>>109104534The one that doesn't look like a butthole.
>>109104534The only AI I use is Gemini integrated with YouTube, mainly for finding timestamps of specific moments in a video
>>109104578>Hw says on the normie tech boardKwab
Which local model is most similar to grok?
>>109105880Yeh. Like, totally. Btichgoy faggot ahh reddit like spacer
>>109105140Yeah no it's gonna just gaslight me before I get bored with it and turn back on night shift nurses or something
>>109104534Grok and Gemini
>>109104534None of them. I am white and don't need a digital liar to think for me. That's for the jeets.
chatgpt for execution, claude for planning, gemini for deep research, and z.ai for long horizon. i just wish there was a way to have them all running at the same time working together to accomplish this, rather than manually feed thee output of claude into gpt for example.
Cursor Enterprise that my job pays for
Summer Dragon
>>109104534Gemini. Grok has insane limits so its unusable Gpt is actually retarded and can't remember things you said in the same conversation.Haven't tried the others.
>>109104534Claude / GPT on PC + Brave and Duck quick summary. Gemini on Phone.
>>109107088Expand on this please.>>109105974Based reviewer. Thanks.
>>109104534gemini but only really ever ask about loonux and related topicsdon't really have a use for the chat besides that 2bh. it lies and makes up enough stuff that I couldn't trust it for anything I don't have some prior knowledge about
Qwen, Deepseek and Gemini, claude free limits are so low I mostly pretend it doesn't exist, Grok for image edits, what I actually use the most is Perplexity until my free subscription runs out lol
>>109104534gemini and chatgptgrok and claude have too many limits on the free tier
>>109104534Mostly Claude, it's absolutely insane how good it is with Linux. But it makes sense, command line werks with any Distro and LLMs love text and code. They were all trained on Linux and open source code.Even Windows help is decent, as long as it's not the ever changing GUI. Claude literally changed my life. Suddenly I'm a Linux, Bash and Python god.
>>109107882>Claude literally changed my life. Suddenly I'm a Linux, Bash and Python god.It certainly makes running my own server and self hosting stuff way better. I'm very glad I didn't have to experience the absolute pain in my ass some of this stuff would have been prior to AI.
>>109104534Lumo
>>109104534Free AI has already solved 99.9% of problems in this world.
>>109104534Gemini for random shallow research on the spot (only model I have on mobile)ChatGPT for news/market, deep research and brainstorming/ideationOpus 4.8 and Compose 2.5 for codingOllama (Gemma4 31B) for basic NLP tasks I couldn’t bother to pay the tokens for.>>109104553Social Media research and fact-checking.
>>109104534Duck.ai, because all of this "AI" shit is nothing but a passing curiosity and meaningless game to me.
>>109104534>Which one do you use?I don't use any of them, because I'm not a retarded zoomer faggot.
>>109104534I don't use AI on my phone
>>109104553For stupid shit I don’t take too seriously. Also for looking up things on my car’s map.
>>109104534I don't use any AI app, but I find the AI results from Google useful sometimes, particularly the links to things it extracted info from. >picScreenshot? Too simple.Photo of a screen? Too mundane.Photo of a cracked screen? There we go.Fucking WHY?
>>109105737>>109105745why do you like gemma4? it doesn't stick to instructions ime
>>109104553Porn. That's basically it. Of course, you can use any old local model for the same purpose and not submit your sickfuck fantasies to Elon Musk. It's dumber that ChatGPT and Claude, and its lack of censorship is counterbalanced by its equal lack of privacy. I suppose Grok has no use case.
>>109104534ChatGPT, because despite being run by a literal cocksucking faggot, it is somehow less faggy than Claude.
What's the best AI for general conversation? I'm tired of asking a simple question, or talking about my day and getting a 2000 word essay with 30 different bullet points.
I used Gemini once to fix my Firefox UI css.That is all.
>>109104534remember when artsisters thought inserting random bullshits into their art would "poison" the model?
>>109108993Claude has the shortest answers I think
>random bullshits
>>109104534Claude Max (5x) and SuperGrok Claude for technical questions and coding. Grok for current events, social media shit, and as a voice assistant in my car. Might upgrade to 20x Max soon. Opus is a fairly competent reverse engineering partner for the autistic shit I do.
Mistral for daily use, Claude for coding
>>109104553it's trivially easy to jailbreak compared to the others
>>109104534None.
>>109105411Maybe consider paying ironically?
>>109104534gpt because it gens images and its the easiest one to use and cheapest monthly. grok is fucking dog shit and gemini is google shit copilot is microshit. maybe I should try deepseek. Also claude is probably way more money and far more restrictive.
>>109104534brain
>>109104534>Qwen 3.6 27BNo matter how much better they are, I refuse to pay for an LLM. There's too much setup on my end to just make these things useable, that I can't reconcile paying for it.
>>109108993Skill issue. Write instructions for brevity.
>>109108993literally Sesame is built for this
>>109105782then what do you suggest?
>>109104553Grok doesn't BS and constantly ask annoyingly useless follow-up questions. It's far and away the best balance of useful and pleasant to use.
>>109105880Still not worse than Gemini's personality.
>>109104534i only have extensive xp with claude and chatgpt but went back to only using chatgpt. theyre both becoming increasingly annoying but fable 5 was literally unbearable. i wanted to elbow that garbage into a pulp and piss and shit on its remains. just thinking about itn makes me heated
>>109109411how good is mistral is it better then gpt?
>>109104553best behaved with my custom prompt, doesn't have an "ethical override" imperative
>>109104534Which one is which AI?for me:> grok is eddyButter toast motherfuckers.
>>109110168This. For general use cases, ie look up this, what do think about that, write me a one liner to do xyz, or simple script, Grok is the best. Especially on concise mode. GPT is the most insufferable of all the AIs, unironically (literally) makes doing anything with it a complete time suck, its infuriating. My least favorite AI. Claude (was) scary good, probably the best, but theyve began knee capping it, and Ive seen how these things go, sooo unless they turn it around and stop destroying their current models, idk where to go next.
>>109104553will the grok still generate child porn images?
recently gemini told me that it's ai model so it tells me what I want, because it started bullshiting on me on second question because I used laughing emoji.They are all fucking garbage, only image and video gen matters, also grok on premium is stripped from full power and scams you to get the highest tier, no idea who the hell unironically use it, they are fucking shit, must check chat gpt for questions and problem solving
>>109111057Why are you prompting with fucking emojis, retard?
codex
>>109104534ClaudeEverything else is shit especially Gemini.
>>109111273fpbp
>>109104553No other model can give real answer on the state of society.
>>109111371Basically for any real fact finding mission. Others are too censored and plug in their own weird politics to give you DEI answers for anything controversial, or any subject that DEI touches.
>>109111057>grok on premium is stripped from full power and scams you to get the highest tier,I use multiple AIs and frequently get the best results from Grok on mere premium. Rarely feel like I need to upgrade (or even ask it to think harder).Generally my usage order goes something like this1. Brave AI Search Summary or Brave Leo (70%, fast, simple, not personalized, citations easy to follow, uses free models)2. Gemini (10% - quickest alternative when brave shits the bed) 3. Grok (18% - Anything requiring more complexity and rigor, or where Gemini is blatantly lying)4. Github Copilot (2% - mostly just for convenient analysis of a github repo)5. ChatGPT (negligible - In theory the best and most accurate model, but I've only needed it a handful of times and have never even been tempted to pay for the higher subscription)I also have a corporate Claude Code account but don't use it for standard queries.
gemini was shrimply better but suddenly webpage started bugging. Now i use https://deepai.org/
>>109104534My own brain because I possess a 150 IQ and don't have to outsource my cognition to silicon valley demons who will soon enslave you. Stay jelly.
>>109104534GPT. I miss Perplexity's Deepseek 1776
>>109111273Why aren't you faggot maggot?
>>109104534Claude Opus mostly. GPT-OSS-20B locally, until I get my dual-GPU setup up and running (lacking the VRAM for bigger models atm).
>>109111346ready to give them your id anon?
