New Delhi:
There is a new child in the big language chatbot block with artificial intelligence, and is threatening to take out the rest of the water. Know Deepseek, developed by a Hangzhou headquarters with a budget fraction (if you think rumors) used to do chatgpt, Gemini, Claude AI and others created by computer laboratories with headquarters in the United States.
And the latest offers: Deepseek V3, a parameter of 671 billion, the ‘mixture of experts’; and Deepseek R1, an advanced reasoning model that uses AI, possibly better than Openai 01, has underlined its status as a possible financial and technological disruptor of heavyweight in this field.
How much of a disruptor is it?
On Monday, this application, The Deepseek V3 Model, is now the best application downloaded in the Apple store in the USA.; Let that sink … A chatbot developed by Chino is now the most unloaded application in the US.
And that interruption, even if it is considered a “potential” at this time, has raised doubts about how well some technological companies in the United States have invested billions committed themselves to the development of AI.
In any case, the quality and profitability of Depseek models have turned this narrative; Even if, in the long term, this particular Chinese model flows, which was developed with a fraction of the financial and technological resources available for companies in the West is a revelation.
Again, how much of disruptor is it?
Well, last month the Depseek creators said Train the V3 model required less than $ 6 million (Although critics say that the addition of previous development stages could boost possible costs north of $ 1 billion) in the computer energy of NVIDIA H800 chips. “Depseek really built OpenAi for $ 5 million? Of course not,” Bernstein analyst, Stacy Rasgon told Reuters.
But the available finance broke down and becomes quite remarkable.
Openai 01 charges $ 15 per million input tokens.
R1 of Deepseek charges $ .55 per million input tokens.
The price, therefore, flys absolutely to the competition.
And, depending on the cases of final use, It is believed that Depseek is between 20 and 50 times more affordable and efficientthan Openai model 01. In fact, the results of the logical reasoning test score are amazing; Deepseek surpasses Chatgpt and Claude AI for seven to 14 percent.
Dev.toA popular online community for software developers said it obtained 92 percent to complete complex tasks and problem solving, compared to 78 percent per GPT-4.
Input tokens, by the way, refer to information units as part of a message or question. These are basically what the model needs to analyze or understand the context of a consultation or instruction.
For the context, it is believed that Operai spends $ 5 billion every year to develop their models.
So, even if Deepseek’s critics (see above) are correct, it is still a fraction of Openi’s costs.
This is translated, as the company’s head, Sam Altman, pointed out in significantly improved computer capabilities, but for the Deepseek model to deliver at least that great processing power in its relatively reduced budget is an rise of eyebrows.
And Altman recognized him, calling the “very impressive” model.
The Google Chief, Slándar Pichai, was a step further, telling CNBC in Davos: “I think we should get the development of China, very, very serious.” And the president of the United States, Donald Trump, has sounded a call to “arouse” the giants of American technology and computing.
And there are the hundreds of billions of dollars that US companies have lost in the midst of a defeat this week in technological actions; The Nvidia chips manufacturer, for example, lost more than $ 600 billion and the Nasdaq index rich in technology technology ended Monday by more than three percent, with the unwanted possibility of a greater fall based on the giants of AI and expected Microsoft profits reports.
For the context, Meta and Microsoft have their own AI models, at the forefront of which are flames and co -pilots; The first is a LLM that was first launched in February 2023 and the second is now an integrated feature in several Microsoft 365 applications, such as MS Word and Excel.
Although it is not possible, at the same technological level as OpenAi or Chatgpt, Meta and MS have invested billions in AI and LLM projects, both in the United States and abroad. For example, some analysts believe that large companies in the US cloud. UU. Garraran $ 250 billion this year only in AI infrastructure.
But what really makes Depseek special is more than cost and technology.
It is that, unlike its competitors, it is genuinely open source.
He The R1 code is completely open to the public under the MIT licensewhich is a permissive software license that allows users to use, modify and distribute software with few restrictions.
This means that you can download it, use it commercially without rates, change its architecture and integrate it into any of its existing systems.
Deepseek is also faster than GPT 4, more practical and, according to many experts, even understands regional idioms and cultural contexts better than their western counterparts.
There is much more to consider.
How, for example, Depseek affects diplomatic and military ties between China and the United States (and India too, actually), and what are ethical problems with open source AI models?
But what is undeniable is that Deepseek from China is a disruptive, in financial and technological terms. And experts believe that China has now jumped, 18 to six months behind the latest generation models developed in the United States.
With contributions from agencies
NDTV is now available on WhatsApp channels. Click the link To get the latest NDTV updates on your chat.
By
