
The same performance claim much cheaper
Deepseek, which attracts attention with its open-source Deepseek V3 model, which requires much less processing power than its competitors and developed at a cost of less than $ 6 million, claims to offer the same performance with its competitors such as Claude-3.5 and GPT-4O.
Deepseek, which is available as web, application and API, has been developed as a competitor to artificial intelligence assistants such as Chatgpt with coding features such as content creation and research. The company also launched Deepseek R1, a rival of OpenAI’s “Thinking” artificial intelligence model O1 last week.
Deepseek’s low -cost artificial intelligence model led to a shares in global technology, and caused shares of companies such as Nvidia, Tesla, Amazon and Meta to decrease.
The company’s headquarters is located in Hangzhou, China, and was founded in 2023 by Liang Wenfeng, who started the Hedge fund that supports Deepseek. In order to improve the technology of the company, it is stated that it stocks the NVIDIA A100 chips before the US export ban and matched them with less power chips that they can currently import.