[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / s / t / u / v / vg / vm / vmg / vr / vrpg / vst / w / wg] [i / ic] [r9k / s4s / vip] [cm / hm / lgbt / y] [3 / aco / adv / an / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / pw / qst / sci / soc / sp / tg / toy / trv / tv / vp / vt / wsg / wsr / x / xs] [Settings] [Search] [Mobile] [Home]
Board
▼ Settings Mobile Home
/g/ - Technology

Name
Options
Comment
Verification
4chan Pass users can bypass this verification. [Learn More] [Login]
File
  • Please read the Rules and FAQ before posting.
  • You may highlight syntax and preserve whitespace by using [code] tags.

08/21/20New boards added: /vrpg/, /vmg/, /vst/ and /vm/
05/04/17New trial board added: /bant/ - International/Random
10/04/16New board for 4chan Pass users: /vip/ - Very Important Posts
[Hide] [Show All]


[Advertise on 4chan]


Redpill me on tokenomics /g/, if it currently costs 10$ to reveng photoshop instead of paying 100$ license, will ppl be building 10 photoshops per year in a year when the price drops to 1$ per photoshop clone? Are we going to create a photoshop of the day to keep the consumption up? Or are we shitting ourselves with these predictions?
>>
Really, neither tokens nor money measure any real value, and the problem is linked until it isn't.
>>
>>110179812
>it currently costs 10$ to reveng photoshop
it costs way more than that, even with subsidized llm subscriptions.
>>
>>110179889
ypu can fucking do it for free on a gaming pc (muh 128gb ram finally pays off see strata)
>>
>>110179889
kek, switch to deepseek 4.1, oopsie 10x cheaper already
>>
>>110179912
>>110179957
prove it. vibe a photoshop clone and post it. post your token usage along with it and let's compute the costs either in electricity for your gaming pc, or cloud token costs, along with % feature coverage.
>>
>>110179957
the guy is clearly AI doomer and tokens will start to cost significantly more, lucky anthropic even tho all the lines seem to converge the other way, time to buy anthropic
>>
>>110179972
dude you can put ds4.1 in harness and you will get to artcraft level of repro eventually, tokens are going dirt cheap, where's the demand? N-no you need to use claude and they stopped REA, so photoshop is safe kek
>>
you can do anything you need with photoshop CS5
it's a shame they fucked up the UI with CS6 because that version introduced some cool new features involving blurring, but whatever, too unusable. CS5 is the way to go.
>>
>>110180002
90% of the way there gets you foss gimp, having a ps copy on your pc so ai can verify visually by screenshotting is all you need, literally qwen3.8 27b can do it with some patience kek
>>
>>110179996
ok, what are you waiting for? clearly you have some experience and it sounds so easy. i'm just a stupid snailcat, I don't have the knowhow. it's only 10 bucks according to OP, maybe even cheaper according to you. go ahead and prove me wrong.
>>
>>110180040
oh no the tokens just went up in price... Stop reversing photoship you guise, the ai is dying!!!
oh wait glm 5.4 is right around the corner and another 10x lower price, can you point to the 'price increase' in anthropic s1?
>>
>>110179812

You're not going trust some activist or portfolioLet to develop the software, especially enterprise stuff. As long as there are devs that dedicate themselves to a project, they're always going to win, especially now.

It leads to digital communism and IMO one of the more central questions is whether corporations and governments are going to let it happen. It obviously completely fucks with IP law and incentives on openness.

There is going to be a larger premium on trust and transparency, because the models themselves might be a bad actor.

I suppose value is just going to shift even more into the doing as relative to design, licensing brands etc. Especially when LLM in themselves reveals exactly how much they are getting cucked, in a consumerist world where everyone is kept dumb, at least in the west.
>>
>>110179812
people will hopefully find better use for their tokens than just recreate ps ad infinitum
that said, the economics are very profitable, with DS reporting they break even on their gpus before even one year, and it's one of the cheapest providers of inference
>>
>>110180062
I dunno what you're talking about. I don't give a shit if you reverse photoshop, in fact I would welcome it. do you need a link to a cracked photoshop to work from? if it's so easy to set up a go ahead and show me it costs "10x less" than $10 = $1 to clone photoshop. what's stopping you? why do you keep saying eventually, and to wait for another model? why can't you do it right now? is $1 too expensive or something? it should be so easy, just set it up and clone it. I'm just a stupid luddite, show me how it's done
>>
>>110180120
look retard, money conscious ppl have been using gimp, one guy with 1T ram can reveng anything, where is the NEED for more tokens if one guy with a server can serve 100k ppl, unless ppl will want to reveng photoshop of the day, the token consumption is not there, muh anthropic is going to the mooon, on what tokenomics? Because surely not on sw prices
>>
>>110180111
>>110180106
if code value goes to zero, what is your product that generates a lot of zero worth worth??
>>
>>110180002
CS6 has almost an identical UI to CS4, and is again almost identical to the current UI (except that has ads and popups everywhere)
>>
>>110180143
isn't the guy with 1T ram named deepseek and glm and those others you keep namedropping? that super ultra cheap model. I said nothing about anthropic or anything going to the moon, I am talking about right now. OP is asserting it currently costs $10 in tokens to reveng photoshop. show me that assertion is true right now because otherwise the entire premise is shot before we extrapolate anything. surely it should be as simple as you keep trying to make it out to be.
>>
>>110179812
Would be hilarious if the AI companies are selling cached photoshops for $10 a month and pocketing the rest.
>>
>>110180171
you can literally install deepseek harness and run on your own 5090 qwen3.8 or qwen flash next per strata for 0 cost and it will develop same features as artcraft (copying 90% from gimp just like artcraft, with screenshot feature, it is literally worthless or at cost of electricity kek)
>>
Photoshop and tokens are both useless.

Neither are worth anything, you can't touch photoshop and you can't eat it either.
>>
>>110180164
products that have sw and hw vertically integrated
services where the client is easily replaceable/reverse engineered but not the content/server
basically anything with a moat that doesn't just depend on 1s and 0s
>>
>>110180164
Well I don't know why replied to me. Like I said there is still trust and transparency. Who are you going to hire to validate the product? You can't have and don't want an everchanging program or marketplace due to the overhead.

However if you're coming from /biz/ and are totally profitbrained. I mean cybersecurity is becoming an arms race, and you are probably going to want some really good human experts on that. Again, regulations and corporation lobbying is the real thing you want to be looking at.

If I was a powerful tech company right now competing against open, I would be fuck up everything and then call for stability with my products. Again look at corporations and government. If anti-circmunvention laws imposed by US goes away then we are probably going towards digital freedom and free market cyberpunk/utopia.
>>
>>110180191
And funnier yet when Photoshop gets a cut too. And then when they find out you’re not really poor they get you for even more!
Whew how deep does it go?
>>
>>110180208
I don't have a 5090 or a deepseek subscription or qwen or any of that. can you lend me a hand? I'm just asking for a dollar. a dollar to prove me wrong for ever and ever hereafter. everyone could point at your example and say look, this guy cloned photoshop on his 5090! (forget that the card cost $10k thats ok everyone has a 5090 already) and we'd have some nice hard numbers to settle the score instead of speculating for the jillionth thread in a row with nothing to show for it despite the supposed ubiquity of the tools available. show me and everyone for the low cost of 1/8 of a grande pumpkin spice latte I could have my own photoshop
>>
>>110180269
you can buy a laptop 5090 for like 3k$, you'll sell surely more photoshop subscriptions hell yoir ai agent can run reverse scams and sell photoshop licences in nigeria, it's all endless sw profit after all like anthropic sells it is surely nierians can buy your nigger photoshop clones in the thlusands per hour jist like anthropic can produce, endless profit
>>
>>110180002

main issue with the last 2 CS-series Photoshops is that their UIs do not properly scale for 2k and 4k resolutions. so you're using a microscope to click on menus.
>>
>>110180285
endless tokens, wooohoo, we to the moon
>>
>>110179812
This is tokenomics:

Your money: 0

Israel's money: all of the money in the world
>>
>>110180285
figures. nothing to show for all that talk, yet again.
>>
>>110180316
gotta invest to earn bro, buy ai stocks and you will be billionaire tomorrow, srsly if youcan't afford 5090 what are we even talking bro
>>
>>110180215
There you go: and regardless of what we invent, you cannot expect a similar shared dependency from any AI.
>>
>>110180316
get your mom to remortgage and have a 5090 earn you your next ferrari rewriing photoslop
>>
>>110180331
and here you are too, supposedly with a 5090 running dsv41 with nothing to show for it, can't put your money where your mouth is. pathetic.
>>
>>110180358
stay poor, my next photoslop already earned me 20k dueing this exchange
>>
>>110179812
>tokenomics
i never heard this word outside of /biz/
>>
>>110180506
how tf do you value ai then?
>>
>>110180208
Is a 5090 faster than a m5 ultra max studio for something like this?
>>
>>110180563
yeah, but you're limited to qwen3.8/qwen flash next (with strata), moe models use the extra speed of vram a lot (or if you can fit the whole model in vram like 27b), otherwise macs are on par with the next nvidia grift: dgx/rtx spark, muh unified memory
>>
>>110180563
if you can fit whole model in vram something like qwen3.8 ninfer qat will run 80 tok/s on 5090 laptop vs 30 on spark (20-40 on mac), but if you want to run something like ds4.1 you need a shitton of ram an yway so maybe a 128 spark can use 2q to run 10t/s and no way to squeeze anything above 5t/s on 5090 laptop even with 64gb ram as it will have to use ssd speeds and ram speed takes over, it really depends on the model and architecture
>>
>>110180643
(then again spark 128gb is priced at like 9k$, so yeah, fuck that shit for now), chinks will come up with better shit in a week or two anyway



[Advertise on 4chan]

Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.