DeepSeek, China's AI model: News & Discussion

[JsCh]

India can’t be left behind in AI. @Krutrim has accelerated efforts to develop world class AI. As first step, our cloud now has DeepSeek models live, hosted on Indian servers. Pricing lowest in the world https://bit.ly/4hxS7wMDetails on our AI lab, SOTA model and research progress, open source drops on 4th Feb!

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

 

[JsCh]

Cloud Giants Offer DeepSeek AI, Restricted by Many Orgs, to Devs​

• By David Ramel
• 01/30/2025

Microsoft, Amazon and Google have all embraced new open-source DeepSeek AI technology from a Chinese company, despite its usage being banned by many organizations.

Cloud Giants Offer DeepSeek AI, Restricted by Many Orgs, to Devs -- Virtualization Review

Microsoft, Amazon and Google have all embraced new open-source DeepSeek AI technology from a Chinese company, despite its usage being banned by many organizations due to concerns about data security, privacy, compliance, and national security risks.

virtualizationreview.com

DeepSeek Now In Perplexity’s AI Search—U.S. AI Dominance Challenged

China’s DeepSeek begins to integrate with international AI players. Perplexity's AI search gets a supercharge from DeepSeek R1. U.S. global AI dominance is challenged.

www.forbes.com

 

[F-22Raptor]

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

 

[BHAN85]

Thread 'Cyberattacks against DeepSeek escalate with botnets joining, command surging over 100 times: lab'

Jan 31, 2025

Cyberattacks against DeepSeek escalate with botnets joining, command surging over 100 times: lab
By Global Times Published: Jan 30, 2025 06:22 PM

Cyberattacks targeting Chinese AI start-up DeepSeek suddenly escalated on early Thursday with attack commands surging by more than 100 times compared to a previous wave of attacks on Tuesday, the Global Times learned from Chinese cybersecurity firm XLab on Thursday.

The lab said that it observed at least two botnets participating in the attacks on Thursday, launching two waves of assaults.

DeepSeek has been subjected to...

• BHAN85
• Replies: 0
• Forum: World Affairs

Shameful

Western "cybersecurity" services need to start to disclose their stupid and shameful kiddie online actions.

 

[iblini]

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

This thread is called,

"DeepSeek, China's AI model: News & Discussion"​

 

[Vkdindian1]

Ask yourself is ChatGpt going to tell you about all the fukc ups by the US? Are they going to give an objective response on Gaza for example?

That’s true. Good guys are as good as they get based on the audience they cater too.DeepSeek does give wonderful results for most people.
It is encouraging and an eye opener for others to get going and create something at fraction of the cost compared to the western world.

 

[Hendarto]

This is what would happen if a government prioritized education and made it accessible to all by lowering the cost and building more universities

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

 

[Aziqbal]

Deepseek wiped $1 trillion off the USA stock exchange

China did in 24 hours what 20 years of wars didn't do

GO CHINA!!!

 

[Anatolian]

First thing I asked was about what happened in Tiananman square. Immediatly failed, so highly doubt this will be popular or widely used.

 

[JsCh]

On February 1, the Silicon Cloud team and Huawei Cloud launched the large-scale cloud service platform SiliconCloud based on the DeepSeek-V3 and DeepSeek-R1 of Huawei Cloud Ascend cloud service.

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

This is wild! DeepSeek R1 and V3 are now available in Windsurf.Fully hosted on Western servers.

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

 

[JsCh]

And on the hardware side,

BREAKING: China Company SiliconFlow & Huawei Cloud Launch DeepSeek R1 Inference on Non-Nvidia Chips. SiliconFlow and Huawei Cloud have unveiled the DeepSeek R1 inference service on Huawei’s Ascend Cloud, offering high-performance AI at the same quality and price as the official DeepSeek API. This breakthrough empowers developers to run advanced AI models on alternative chip architectures, freeing them from Nvidia dependency.

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

Deploying Deepseek R1 distilled "reasoning" models on AMD Ryzen AI processors and Radeon graphics cards is incredibly easy and available now through @lmstudio. See how to quickly get them up and running!

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

The era of better #AI is here — thanks to the power of an open ecosystem.The entire Intel product portfolio, along with our robust suite of developer resources, already supports hundreds of thousands of open-source AI models, including @deepseek_ai.

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

Securely experiment and build your own specialized agents, as the 671-billion-parameter DeepSeek-R1 model is now available as an NVIDIA NIM microservice in preview on http://build.nvidia.com. Learn more https://nvda.ws/4grQaBq

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

 

[UKBengali]

Deepseek wiped $1 trillion off the USA stock exchange

China did in 24 hours what 20 years of wars didn't do

GO CHINA!!!

Nvidia is now 15% down and it will get worse for them as the Chinese reveal more "surprises" with their far more efficient training models over the course of this year.

Deepseek are apparently now training using Huawei Ascend 910C GPUs which are roughly comparable in performance to Nvidia H100 from 2022. They are cheaper and more power efficient compared to H100.

 

[BananaRepublic]

Nvidia is now 15% down and it will get worse for them as the Chinese reveal more "surprises" with their far more efficient training models over the course of this year.

Deepseek are apparently now training using Huawei Ascend 910C GPUs which are roughly comparable in performance to Nvidia H100 from 2022. They are cheaper and more power efficient compared to H100.

Tech and finance are the only areas where the US has supremacy over the Chinese.

When China catches up and then surpasses USA on chips that will be a pivotal moment for US hegemony.

That will then change the dynamics on finance as lot of Arab money is going into US tech. And ofcourse oil being priced in Dollar.

Very soon 99% of devices will run on Chines chips - that will itself be a game changer.

Leaving US chips for niche stuff like AI training.

 

[UKBengali]

Tech and finance are the only areas where the US has supremacy over the Chinese.

When China catches up and then surpasses USA on chips that will be a pivotal moment for US hegemony.

That will then change the dynamics on finance as lot of Arab money is going into US tech. And ofcourse oil being priced in Dollar.

Very soon 99% of devices will run on Chines chips - that will itself be a game changer.

Leaving US chips for niche stuff like AI training.

Even if US lifts all restrictions on China, they won't ever use US GPUs ever again.

They have simply no need anymore as Huawei/SMIC now have viable alternatives.

Nvidia should never have been a trillion dollar valuation as it is fundamentally a company that makes gaming/creator GPUs. They will soon go back to their bread and butter business.

AI is massively overhyped and AGI is nothing like what we are seeing now with these LLMs.
- we know they just spit out what they have been trained on and cannot correct themselves and "learn" from conversations like a human can. Sam Altman has now had to roll back on his claims of AGI this year as no one has any idea how to make a computer "think" like a human.

 

[BananaRepublic]

Even if US lifts all restrictions on China, they won't ever use US GPUs ever again.

They have simply no need anymore as Huawei/SMIC now have viable alternatives.

Nvidia should never have been a trillion dollar valuation as it is fundamentally a company that makes gaming/creator GPUs. They will soon go back to their bread and butter business.

AI is massively overhyped and AGI is nothing like what we are seeing now with these LLMs.
- we know they just spit out what they have been trained on and cannot correct themselves and "learn" from conversations like a human can. Sam Altman has now had to roll back on his claims of AGI this year as no one has any idea how to make a computer "think" like a human.

Having trained an AI - I know exactly what you mean.

The output is what you feed it and how you feed it.

You also have to explicitly tell it not to do certain things e.g. do not recommend a rival’s services lol

I use chatgpt and copilot for coding.

ChatGpt needs 3-4 goes to produce compilable code.

And copilot is a glorified predictor.

However, both are brilliant productivity tools and replaced stackoverflow for me and YouTube for self help.