@gigaqian: DeepSeek just released an inference architecture upgrade that makes compatible AI models up to 400% faster! It’s called DSpark, and it’s already making the recent DeepSeek models more efficient and much cheaper to use. This is showing how the new focus in the AI race is about inference speed. DeepSeek is showing us how to compete the open source way! #breakingnews #deepseek #aitechnology #ainews #gigaqian
Love deepseek but any news on a new mimo from Xiaomi?
2026-06-29 06:13:53
0
Ln :
Does it solve the water crisis with ai data centers
2026-06-29 14:26:52
27
George Patience242 :
wait till you get a load of DeepSeek Cephalopod 🐙...😳
2026-07-08 07:17:50
0
ChasingGnosis :
I'm months from understanding all this but I listen anyway in hopes osmosis works.
2026-06-29 00:08:04
34
Exhibition Gallery :
DeepSeek rocks especially their computer vision API. 😁✌️
2026-06-29 02:04:50
12
Lucian :
have you heard of recursive mas ?
2026-06-29 16:01:01
1
Chuck Bartowski :
Brother, sometimes I think you're AI... You really have a finger on the pulse of the bleeding edge here in AI development. Are you in the industry?
2026-06-28 23:57:12
13
R Ulises Mendoza :
could you refer the paper to check on this? thanks. Great content!
2026-06-29 01:57:01
3
Roger Hahn :
Makes so much sense. Back in the day, I used Perl to code markov matrices for predicting next nucleotide in DNA coding regions. Next token prediction for inference is a no brainer app for Markov models. Doh! Curious if others thought of this too.
2026-06-29 04:39:06
3
_LucasDucas_ :
So if I were to buy Nvidias dgx spark and stack it with another I wouldn’t be able to run this new model of deepseek? Or are you saying it would be cheaper and more beneficial for me to just use huggingface
2026-07-01 13:39:02
0
Backstage Party :
You rock dude!
2026-06-29 03:39:12
10
Wahala :
Further tech
2026-06-29 05:15:32
1
brainrottingg :
Too confusing
2026-06-29 03:59:28
0
Utopidy :
I run deepSeek locally on my desktop. Yes it is an older version, but still suffices for my needs. It's great!
2026-07-28 15:08:48
1
flux :
where did it come from.
2026-06-28 23:54:34
1
objectivelycorrect_ :
Music choice goes crazy lol
2026-06-30 00:17:59
1
The Realest :
First they made models with selective parameters(35B at a time), then making AI write whole sentences at once, instead of 1 word, and now this... I love this company !!
2026-06-29 15:57:00
4
daphne_the_boxer :
First
2026-06-29 01:14:40
1
Vadar :
Thank you
2026-06-28 23:51:25
2
fundamentally incremental :
they're so efficient 😫 god! that's what I love about it.
2026-06-29 15:47:35
1
Jamaican :
My goodness
2026-06-29 00:13:11
1
Bumi Kita :
i just ask deepseek how to implement dspark. it said: " do not have to lift a finger. just keep using the api...".
2026-06-29 10:16:59
1
Irongate :
you are already behind. ai can show a pic of thousands of words or write a whole book at once not word by word 😂
2026-06-29 01:02:14
1
Venkata.M :
speculative decoding has been here for a while...
2026-06-29 13:21:23
0
Laura 🇨🇦 Author-artist-poet :
DeepSeek is amazing
2026-07-13 01:08:34
0
To see more videos from user @gigaqian, please go to the Tikwm
homepage.