Artificial Intelligence thread

9dashline

Captain
Registered Member
Board fight and talent drain to other startups is what has happened.
Since I paying $200 mo I tested it, according to its alleged strength, to have it write me a short fictional story, I gave it same test I gave R1 and o1-pro, and yet its 4.5 is worse of the three. Also unable to output long tokens unlike o1 pro or deep research. And its knowledge cutoff of a lot of recent tech is still back from 2023

Please, Log in or Register to view URLs content!
 

tphuang

Lieutenant General
Staff member
Super Moderator
VIP Professional
Registered Member
Please, Log in or Register to view URLs content!

China Telecom's 息壤 platform currently has 22 EFLOPS and if including 3rd party compute, it has 27 EFLOPS. Not sure if this is FP16 or FP64.

It's providing DeepSeek R1-V3 to customers like Sinopec, China Railway Material Group, Dongguan gov't and digital management and other government entities.

Please, Log in or Register to view URLs content!

Alibaba's B2B platform has AI search function Accio using DeepSeek's reasoning model

Please, Log in or Register to view URLs content!

Baidu has fully onboarded DeepSeek-R1 full blooded version. Customer in web PC and apps can all be using it.

Please, Log in or Register to view URLs content!

Baidu Library and Baidu Cloud have all onboarded DeepSeek-R1 fully version
 

Eventine

Junior Member
Registered Member
New start up from the founders of Oculus, Ubiquity6, and former Meta Reality Labs focusing on direct voice-to-voice AI chat (as opposed to voice-to-text / text-to-voice), challenging Open AI and Grok 3.

They've put up a demo here:
Please, Log in or Register to view URLs content!
, it seems quite impressive. They promise to open source the model, which is smart considering what's been happening. The start up is backed financially by Marc Andreessen among others. Presuming their open source version isn't crippled, this could lead to wide adoption; too bad a Chinese lab didn't get to it first.
 
Top