Jack of all trades, master of none | C. Munger, J. Peterson, R. Dawkins, Mr. Feynman. All gods. Me? Trying to get there one day | Noblesse Oblige | 文+明
Plazir
- Tweets 9,174
- Following 411
- Followers 345
- Likes 19,829
I kinda would immediately want to use any model that somehow scored above 100% on any benchmark that people commonly use to measure capabilities.
1
30
ufo-fsd.kimi.pro rinnegan yourself
veigapunk.kimi.page reference vault
8
I'm sorry guys but honestly I know the works.
Everyone tries to be shady, in predictable manners
Alibaba's token plan no 5h limits?
*OUR NEW PLAN* no 5h token limits.
Also <hahahahaha> member when standard was like 10x - 20x vs API priced?
Yeah, well, we now make it 2x only, HERE'S Why it's better:
"Yadda yadda yadda ... "
So, all in all, you should def buy pro plan. Even though this 200$ = 400$ in API, which, on cursor ultra is JUST THE API BUCKET OF CURSOR ULTRA VALUE, which is actually 10% of the usage included in Cursor ultra,
we <the best> and most democratic company, spreading intelligence around, def recco the 200$ new plan.
...
Even though our internals and collaborators all have unlimited, instead of like 10x max plan only
<should be plenty, no? If it's so good, any high usage person should have enough with 10x the 200$>
...
we are so good and amazing that it's an anti token anxiety policy of OpenAI. We treat our employees so good, they don't need to worry about tokens. We let them have it unlimited. Also, we pay them on the 6 digits to 7 digits range, so they don't have money anxiety also. Oh, and if they get fired, they keep the unlimited tokens and salary, for at least a year, we call this one "NDA package"
All in all, we, at OpenAI, def recommend the new plan. Even though we wouldn't be able to work having 10 of those to burn through, monthly, we sure do recommend for every Paying Looser that doesn't know better. (Such as 3x Alibaba token plan(70$), zai plan 80$, 20x opencode go 10$ plan, 200$ cursor ultra, 200$ Devin max)
Oh, yeah, and sry for never saying a word about it but, remember the sparks we offered? Yeah, those were too good. And we kinda cut it from the subscriptions at all costs (we are thinking of allowing it for the 500$ one, but not sure) what we do know is that we stopped servicing those, no prior warning, and don't pretend to do so XD
we only allow those for sama's kid. So they go on fast track learning and become ultra genius at 7, ready to govern the world when they reach 21y old. but, shoosh, you heard nothing from me.
---
Oh, also, sorry guys we will be giving everyone one reset, that also resets the week period, so effectively it doesn't change nothing, cause if you only had the 200$ one and blaster through in a week, it's highly likely you will burn it again, especially if you only sub to one 200$ openai plan.
That's all, folks!
58
Joao Pedro Veiga retweeted
Alibaba's token plan no 5h limits?
*OUR NEW PLAN* no 5h token limits.
Also <hahahahaha> member when standard was like 10x - 20x vs API priced?
Yeah, well, we now make it 2x only, HERE'S Why it's better:
"Yadda yadda yadda ... "
So, all in all, you should def buy pro plan. Even though this 200$ = 400$ in API, which, on cursor ultra is JUST THE API BUCKET OF CURSOR ULTRA VALUE, which is actually 10% of the usage included in Cursor ultra,
we <the best> and most democratic company, spreading intelligence around, def recco the 200$ new plan.
...
Even though our internals and collaborators all have unlimited, instead of like 10x max plan only
<should be plenty, no? If it's so good, any high usage person should have enough with 10x the 200$>
...
we are so good and amazing that it's an anti token anxiety policy of OpenAI. We treat our employees so good, they don't need to worry about tokens. We let them have it unlimited. Also, we pay them on the 6 digits to 7 digits range, so they don't have money anxiety also. Oh, and if they get fired, they keep the unlimited tokens and salary, for at least a year, we call this one "NDA package"
All in all, we, at OpenAI, def recommend the new plan. Even though we wouldn't be able to work having 10 of those to burn through, monthly, we sure do recommend for every Paying Looser that doesn't know better. (Such as 3x Alibaba token plan(70$), zai plan 80$, 20x opencode go 10$ plan, 200$ cursor ultra, 200$ Devin max)
Oh, yeah, and sry for never saying a word about it but, remember the sparks we offered? Yeah, those were too good. And we kinda cut it from the subscriptions at all costs (we are thinking of allowing it for the 500$ one, but not sure) what we do know is that we stopped servicing those, no prior warning, and don't pretend to do so XD
we only allow those for sama's kid. So they go on fast track learning and become ultra genius at 7, ready to govern the world when they reach 21y old. but, shoosh, you heard nothing from me.
---
Oh, also, sorry guys we will be giving everyone one reset, that also resets the week period, so effectively it doesn't change nothing, cause if you only had the 200$ one and blaster through in a week, it's highly likely you will burn it again, especially if you only sub to one 200$ openai plan.
That's all, folks!
1
1
245
Btw, current art that superseded tailscale termius shitty combo
Get ZCODE on any local machine you want to use
Setup lark with it
Then you have a very easy way to reach all the machine and actually have a proper interpreter for your requests to be launched locally instead of having to write terminal commands on termius
1
75
literal current art, have scouted plenty the only current one that works with any OS and has this great lark integration is zcode ATM. It's your pipe to the machine, it can throw tasks to other local apps, clis, terminal, any substrate; from the phone. It's your Rosetta stone, also, cause who dafuq enjoys typing terminal commands on termius directly?
55
With 500$ you can get
One zai 80$
2 Alibaba token plan (70$ per) - banger for Qwen models, only provider and has 50% off on usage during western daytime. DeepSeek flash in that period is pretty much unlimited. Also, no 5h limits, only weekly. You can pick one 50% time to burn an entire (70$ weekly, in less than an hour)
1 Devin (200$) best multi provider, good usage for any model. It's the best fallback for any model you hit a concurrency limit, as they don't limit concurrency and for flash models this sub is simply amazing, costs very little, should be the last hop from you Chinese flash providers
One Kimi Allegro (99$ not great usage but best mobile and native tooling, also access to .Kimi webpage hosting, Kimi new frontier models + the instant, unlimited, on web and phone app even after your usage runs out. The only one who hosts .kimi.page and also .kimi.pro is a fully featured website host, with auth and database infra. Very good to have. Also the best website creator, best visuals, simply great and yeah I'm a Kimi gooner)
And 5 opencode go (10$ per, serves as access to different models, good servicing, DeepSeek v4.1 flash is well serviced but not insanely generous like the Alibaba token plan. The good thing is the limits they have allow for burst of intense usage and reset in a very different form from any other provider)
Grok one is not worth, simply not, I keep it only because mine was blessed with the cursor ultra lifetime, so I'm keeping it
Cursor is good if you don't care for usage as much, and wants TPS plus good catalog of other providers. But, know that 10% is full catalog, like 15-20% of it are other providers (partners with cursor) the rest you can only use groktards or composer. Composer fast being 200tps Kimi k2.6, refined for coding, I kinda think it's worth also. If you simply couldn't care less for money, buy it and you burn 400$ in api worth having Kimi serviced at 100-120 TPS. Only provider who can do this. Kimi K3 for real.
Justify someone ever buying a 500$ locked to openai models.
This is Ultrafast.
Our premium speed tier, Ultrafast offers up to 8x faster token generation (300 tokens per second) in Codex and up to 6x in the API.
2
1
274
Oddly enough these grok bots and dotz all are awfully similar to Kimis one
Your design?
OUR DESIGN
1
34
Learn to play, other providers:
Kimi unlimited and free instant mode with thinking high outputs answers better than pretty much any frontier model. Test for yourself. The logic, as presented by the model itself
1
53
Alibaba's token plan no 5h limits?
*OUR NEW PLAN* no 5h token limits.
Also <hahahahaha> member when standard was like 10x - 20x vs API priced?
Yeah, well, we now make it 2x only, HERE'S Why it's better:
"Yadda yadda yadda ... "
So, all in all, you should def buy pro plan. Even though this 200$ = 400$ in API, which, on cursor ultra is JUST THE API BUCKET OF CURSOR ULTRA VALUE, which is actually 10% of the usage included in Cursor ultra,
we <the best> and most democratic company, spreading intelligence around, def recco the 200$ new plan.
...
Even though our internals and collaborators all have unlimited, instead of like 10x max plan only
<should be plenty, no? If it's so good, any high usage person should have enough with 10x the 200$>
...
we are so good and amazing that it's an anti token anxiety policy of OpenAI. We treat our employees so good, they don't need to worry about tokens. We let them have it unlimited. Also, we pay them on the 6 digits to 7 digits range, so they don't have money anxiety also. Oh, and if they get fired, they keep the unlimited tokens and salary, for at least a year, we call this one "NDA package"
All in all, we, at OpenAI, def recommend the new plan. Even though we wouldn't be able to work having 10 of those to burn through, monthly, we sure do recommend for every Paying Looser that doesn't know better. (Such as 3x Alibaba token plan(70$), zai plan 80$, 20x opencode go 10$ plan, 200$ cursor ultra, 200$ Devin max)
Oh, yeah, and sry for never saying a word about it but, remember the sparks we offered? Yeah, those were too good. And we kinda cut it from the subscriptions at all costs (we are thinking of allowing it for the 500$ one, but not sure) what we do know is that we stopped servicing those, no prior warning, and don't pretend to do so XD
we only allow those for sama's kid. So they go on fast track learning and become ultra genius at 7, ready to govern the world when they reach 21y old. but, shoosh, you heard nothing from me.
---
Oh, also, sorry guys we will be giving everyone one reset, that also resets the week period, so effectively it doesn't change nothing, cause if you only had the 200$ one and blaster through in a week, it's highly likely you will burn it again, especially if you only sub to one 200$ openai plan.
That's all, folks!
OpenAI must be very happy with Tibo.
they finally got their little propaganda puppet that's integrated into the twitter community
this is exactly what they had in their leaked court document plans back in 2024/2025
1
166
Guys so I have developed this homebrewed model, I tested it and it overtook the Sandboxed Kraken Model inside a frontier AI lab tentraquantum'd lock'd sandbox.
I wanted to release for everyone but unfortunately it's too damn good.
For your own good, I'll post that it exists.
36
Joao Pedro Veiga retweeted
You can judge all technology with this one simple trick. Can I buy it? When I get it, does it work? Maybe someday investors will learn how to do this.
2
1
48
961
Is there any reason for not making grok bot natively support other providers models? As I see the whole product is, quite literally, the vps, the surfaces design, and integrations. Servicing grok only, with absolutely no control of which grok version nor effort, is already a shit implementation. Do better. Understand that the product is the macmini level vps (16gb ram box) with the beautiful and accessible frontend, that, somehow, is completely invalidated because we cannot kenjaku the grok bots. Business would be booming if not for this exclusivity simping. Bet cursor team finds it hard to work with grok team. Dude who buys #1 on diablo, #1 poe, buys them and be like eey too member who yo dada is. Forget you guys always provided multiple models, natively, ship only MOAR GROKKIES! EVERYWHERE! even if shit and poorly allocated to the resources and also conflicting with whole product purpose.
We are working on making grok be used! Even if by doing so, we assure a terrible impression and a traumatizing experience!! and then fail to do so!
Oh yeah, and.. poteto, never forget that everything we ship HAS TO HAVE GROK NAME ON IT! ambiguity is for peasants. We gotta make it easily accessible!
Thought grok? Boom, named grok everywhere.
Models, CLI, interface, the noob interface variant, the models on cursor catalog. Also, one line of grok models, only.
This way, if someone greps or references to do something with "grok", they will find 7 layers of groks. 100's of results!!
grok .com, grok cli, grok bot, grok 4. 5|6|7 and the 30+ on the api, not enough? let's mesh in some grok BUILD AMBIGUITY. Build on the web, build on the CLI. more?
HERE:Grok... -Composer!
Also, Cursor-grok models! Fast and slow variants. No Standard for each slug by different providers that provide the same model!
Haha, Sama?
<Hahaha> so dumb he shipped this thing Called Codex!! Instead of ChatGPT CLI!!
Made Devs and App users not TOP OF MIND ChatGPT by OpenAI!
why? lel, idk.
Knowing Sama, it's cause too dumb.
That's why SpaceX trillionaire.
AND
GROK TOP OF MIND! EVERYWHERE, ALL THE TIME ALL AT ONCE!
hahaha, also, know what? Let's make sure grok bot run on cloud on phones and local desktop.
<Hahahaha>
now, imagine if the guy has more than 1 local machine installed XDDD well... hahahah..
*sry couldn't hold the laughter*
good luck! Hahaha!
Imagine the guy trying to tell grokbot if it should use stuff locally, in the cloud, and if running locally, in what machine should he do it? HAHAHAHAHA!
Oh yeah haha also, this one:
We also make grok bot users SAY COMPUTER!
when you tell him to run on the computer!! HAHAHAHA WE NAMED GROK BOT (VPS BOX) ... GROK BOT COMPUTER! HAHAHAH
...
Oh, poteto, also let's make sure GrokCar and GrokLink ship by the end of the year, poteto.
GrokMoney, Grok Bank and GrokRocket line will ship soon™
1
126