Now starting to see why Hugging Face was hacked. 😂
2026-10-20 13:24:02
35
Socrates :
It still costs a lot of money to run the large open models. You need premium GPUs, lots of memory, fast storage and lots of electricity. Not to mention cooling and backup power.
2026-10-20 14:18:18
3
Return to Your Self :
If the frontier labs win, the wealth gap gets worse because 40% of ALL VC MONEY is in frontier labs. But if they lose, it destroys the economy because 40% OF ALL VC MONEY IS IN THE FRONTIER LABS. Our only option is to use JUST barely enough frontier models to keep them in business. But we all have to HALT usage the second they see one single cent of profit. Gonna be super tedious but it’s honest work
2026-10-20 18:56:04
2
user83324586 :
it is just the beginning, AI will be one of the main components in military of the future, so if ai companies run out of arguments for civilian use investments, they would easily raise money from gargantuan military budget
2026-10-20 20:31:59
1
getklarted :
The Nvidia Spark range just dropped also
2026-10-20 22:12:38
0
Peter :
not OpenSource, open weights
2026-10-20 21:51:02
1
hjkkbfhhj :
Yeah but they don't have enought compute to train models that are as good as the ones of Anthropic and Open AI
2026-10-20 14:23:07
3
Gwee Jia Han :
They are realizing AI as a service is gonna die faster if they develop faster.
2026-10-20 14:01:49
13
Evgeny :
Be specific on use cases. I tested data extraction from documents and images on Qwen - works good. But coding without Claude Opus or similar is waste of time
2026-10-20 14:32:42
0
Summer Madness :
anyone can download the model. running it tho in a useful way will cost 10 to 50k
2026-10-20 16:44:54
1
Steffan Klein :
where are we with natural language coding commonly called Vibe coding though in comparison? I use Claude only because of its perceived advantage here as I don't want to break a really complex solution with many functions and database
2026-10-21 03:48:02
0
michael :
Like literally ChatGPT is 6 lunar you can do with an open model for cheaper
2026-10-20 18:12:22
0
yindogg :
Nah. Those stats dont count usage on Anthropic.
2026-10-20 20:19:15
2
NoomNoo :
I use only open source or weight models. Why would I pay a fee for having the same service? I don't practice code or any programming, and I don't work for high tech companies. This race and its rhetoric is BS for common humans.
2026-10-20 13:20:43
0
eusouramiro :
Music is too loud man..
2026-10-20 15:06:45
0
The Gen❌️ Introvert :
Ai is a race to zero. This coinsides with Elon's theory of Ai will pay for everything. I'm still not sure how that will all work. But, we are watching it play out in real time.
2026-10-20 13:06:07
0
w_lizard97 :
haha I knew it's gonna happen and it's really funny to watch if you know the whole story. 😂
2026-10-20 12:45:45
4
🜲 마우리 🜐 :
And the worst part is that closed weights/ai online is just surveillance
2026-10-20 17:36:31
1
Mohamad El-mahdi ASunna :
Little by little, the fog dissipates, the clouds clear, and the vision becomes sharp and distinct; only then will you know whether Dario was a lion or a donkey.
2026-10-20 14:47:45
1
Janani Subramanian :
that pic of altman and dario is funny
2026-10-20 13:23:07
1
Apple User596258 :
Me too man! 😁These labs are playing with the future of humanity and wants everyone else to pay if it goes south.
2026-10-20 13:16:48
1
Redouane :
I mean look all the frontiers model they secretly pushed lite/luna/light/fast cheap versions that also cut 80% of the price. Now they all became useless if you don't pay, sometimes even if you pay the entry price
2026-10-20 18:24:23
0
Mohito :
The math ain’t mathing….AT&T runs 10’s of billions of tokens and so for them to move coding to open weight models makes sense. But note they are still running on Azure, AWS, etc. So it’s not like they are buying own hardware and hosting open weight models…..
2026-10-20 20:43:03
0
To see more videos from user @parthknowsai, please go to the Tikwm
homepage.