>>110267439
Some clients (based on useragent?) are not allowed to use the non-logged in chat anymore. If you're going to login, might as well use claude, which is 100000x better than logged in chatgpt is.
Chatgpt right now is so terribad that it can't even enumerate things correctly. It will start at 33 and 2 turns later will append new items starting at 125 and such. This is gpt-3.5-tier fuckups.
Claude does not have this problems.

So there is 0 point in using chatgpt for anything anymore.

That said I'm very surprised to see perplexity ticking up, it's totally worthless. Everything in this list can do what perplexity does but better, including chatgpt. And I'm likewise very surprised by claude having such a small market share.

>>110267439
Because Google Gemini and Perplexity increased.

>>110267439
They killed usage

are you telling me copilot and claude have the same market share?

>>110267493
>very surprised by claude having such a small market share.
from people i know, outside 6 devs forced by their managers to use claude, and 4 other hobbyists using it for their projects paying out of their pocket burning through tokens, rest of normies dont know or dont care about claude. like literally they go "what the fuck is claude"

>>110267439
If you see fluctuations like this, you should question whether the numbers are or ever were accurate.

>>110267439
It didn't. The graph does not reflect reality. Either the numbers from September and before are wrong, the October numbers are wrong, or (most likely) both are wrong and all numbers there are made up.
>how do you know
Because I'm not retarded. Usage numbers don't move like that.
Are you the same guy that makes threads every month because Linux marketshare has apparently tripled / halved / 10x'd / gone to zero according to some random graph on the internet?

>>110267564
Yes no one cares about Claude. Also no one pays for AI.

gemini. this thing is fantastic; i barely leave the search page now.

>>110267564
I can see it
aside from tech jobs, no business is paying a gazillion dollars for their employees to slop it up

>>110267439
because there's no reason to use shatgpt when gemini and claude exist

>>110267590
>>110267592
similarly, a decade of chinese new year steam anomalies and people lose their minds every time
some people are just too dumb to get it

>>110267439
fucking perplexity is beating claude? lol wtf

makes sense if gemini also considers ai mode in google search
most people used chatgpt to summarize web searches but now the search engine they've always used does it too

>>110267742
said no one ever

>>110267991
k tard. google zero click results went from <45% to >68% in three years.

https://en.wikipedia.org/wiki/Zero-click_result

What the hell is perplexity and why is it so popular?

>>110267991
I use it because it's the only one that consistently provides links to the sources for all its claims. Half the time it hallucinates shit, but so do all the other LLMs. At least with this one I can check the source and know it's wrong and try a different prompt. Like >>110267877 said, basically just a better web search

>>110267439
Presumably it became less free on its usage, and the rising ones are the ones willing to absorb more losses. Is this what a sustainable industry looks like?

>>110268141
I have never not seen it hallucinate its "sources" AND I constantly have to convince it to give me actual links instead of "naming" a "potential source".

>>110268076
>trusted the clanker for a source
>it just made it up
Way to prove your point, cletus.

>>110268332
why google stock fell last earnings?

>>110268399
Because they are last-in-class, even below the 8b chink models, yet spend all their money on AI slop. Or what, did you believe whatever hallucinations the clanker told you?

>>110267493

Not my experience with ChatGPT. I don't use it to do work that requires me to trust it entirely, i.e., I know when it is wrong or 'hallucinating' and it almost never does, because the work I do with it I could entirely do myself -- it's just helping me get more done faster.

Maybe, just maybe, the people having problems are the one who think AI is a replacement for their own mind, and have entirely outsourced thinking to it only to realize after-the-fact that it can make mistakes.

Personally, OpenAI fumbled hard when they killed GPT-4o/4.1. That model should have been open-sourced or at least left as an option for paying ChatGPT tiers. Now, going from the v5 models to v6, the v6 models are not quite as good as v5 in general. Much less trustworthy.

>>110267742
You really should, because most of the time its BS.

>>110268437
wrong. it was because their zero-click results (gemini) started eating their revenues from advertisement.

>>110267439
I mostly use Gemini, even though I also use Claude at work. Gemini sounds more like a machine rather than a human, which is a good thing. OpenAI and Claude try too hard to be cute, but it's completely counterproductive.

>>110267844
Yeah this can't be right.
>>110268087
It started as an AI-powered search engine. It was alright for a while but it very, very quickly turned to absolute shit and they're now pumping their money in weird marketing stunts. Oh and they have their own Chrome fork if you want them to steal ever more of your data.

>>110268499
>my hallucination machine told me so therefore it must be true

>>110268491
That's normal, you have a sub-65 IQ

>want to know more about a theatrical adaptation of a book
>it's a German play
>I can't speak German
>I ask ChatGAYPD to check it out for me. Also ask about who's playing the lead
>GPT gives me a name but also tells me he's playing a character that doesn't exist at all in the book
>"yes, it's an original character created especially for the play!"
>suddenly realise that the actor it namedropped is American
>point out that I doubt that an American actor would go out of his way to play in a small theatrical German production
>"that's true, but that's because I made everything up, lol"
I wish clankers had necks so I could strangle them.

>>110268561
>my hallucination machine told me

no, they talked about it during the last earnings call. it was on every tech and investing websites.

>>110267493
I dont even know what you guys are on my experience is completely alien from that

>>110268646
Thanks for confirming that you are financially illiterate. Or maybe plain illiterate based on your posts so far.

He said bad ram thing

>>110268578
Authentic experience of talking to a woman

>>110267567
>normies dont know or dont care about claude
their loss. claude is goated

>>110267493
>That said I'm very surprised to see perplexity ticking up, it's totally worthless. Everything in this list can do what perplexity does but better, including chatgpt.
Perplexity does not require an account, you can use it without logging in. It would be number one if it had a better name and service, unfortunately the creators are retarded.

>>110268718
https://arstechnica.com/gadgets/2026/10/google-admits-ai-search-makes-less-money-but-its-fixing-that-with-more-ads/

https://247wallst.com/investing/2026/10/29/google-admits-its-ai-search-makes-less-money-than-regular-search/

zero click results are eating the web. why do you think reddit is going login users only?

File: 1793326913595425.png (170 KB, 1350x1007)

170 KB PNG

>>110267439
go to any gemini/antigravity subreddit and you'll understand why. half of india got a trial for Google AI Pro from their cell provider (Jio).

File: yuu.png (32 KB, 100x378)

32 KB PNG

>>110268854
i bought a few of those Jio indian keys, 18 months for ~$5? one got canceled after a week, some still working.
it's not useful since you can't use argon in agy

>>110267439
Gemini is good enough and can be used on Google without any gay sign in

>>110268087
Joe Rogan shills it

>>110268937
because google makes a distinction between free plans (pixel promo, ISP, student) and actually paid plans (need to spend at least $1). I haven't seen any pro plans for less than $10/month even with discount though. Argon is useless either ways since it burns through tokens like nothing else. You could make a case to have Argon do planning and let it call 3.8 Flash (high) subagents to do actually do shit since Flash is decent now. I think it's probably the best free model if you're using it through AI studio, but obviously it can't do everything through the browser.

>>110268792
Perplexity hallucinates links and quotes far harder than google's search ai though.

>>110268809
>more self-pwn
lmao