Deepsek challenged at only US $ 14? – Thaydayafterai News

Photo of author

By aispaceworld



Fei-Fei Li, Titan of an inventory of an inventory, opened S1-32b-the reasonable forms of economic and technical interpreter. Details on January 31, English, English – its competitors may be US $ 5.6 million.

[Read More: AI Pioneers as Santa: Stuart Russell, Sam Altman & More Celebrate Christmas with X Grok 2]

S1-32B model: New distillation new standards

Li, lead the team from Stanford and University of Washington, breakilled S1k about Alibaba Cloud. Excellent format in reasoning O1-preview by 500% in the problem in 2024% in accurate stimulus by stimulating themselves. Click this leap in the v3 / R1 efficiency of the Deepseek, which, while not on the extensive $ 5.6 million.

R1 of Deepseee, launched on January 20, stunned with low training budget and US Store Store Store asc Store Store Store asc Store Store Store. However, just raz-jozor tag-off-Over-over 100,000 times cheaper in the restriction. Li’s team honors their credit and light hair cuts, which contrary to the best height in Algorithmic and Hardware system.

[Read More: DeepSeek vs. ChatGPT: AI Knowledge Distillation Sparks Efficiency Breakthrough & Ethical Debate]

Cost Outlook: Multiple view

The interpretation of making clouds are expensive for S1-32B model differs across industrial stores. Panitslab.com And the verge Both reports that all prices for S1-32B training is under US $ 50. Tim Kellogg’s BlogHowever, it is estimated that the cost of lower costs about $ 6, even if this will not be in line with normal clouds. Current rental rates from the provider, such as hyperstack, estimated value for the 26 minutes is approximately 26 minutes in accordance with Economic accounts.

[Read More: Can a US$5.6 Million Budget Build a ChatGPT-Level AI? ChatGPT o3-mini-high Says No!]

Fei-Fei Li: Travel of Friendship

Stanford Ai Pioneer Fei-Fei Li, awarded 2025 Queen Elizabeth Receive Award for Engineering, Continuing formatting. Beijing in Beijing in 1976, Li gets refugees to the United States, the Chinese restaurant. She has given a Physics degree from Princeton in 1999 and PhD. In electrical engineering from Caltech in 2009. In 2009, she created a computer vision and determined modern learning methods. Now the Professor Stanford and the World Staters’ co-laboratory, access to understanding of the world’s physical world in travel The world I see.

[Read More: DeepSeek AI Faces Security and Privacy Backlash Amid OpenAI Data Theft Allegations]

The role of Qwen: Open versus elite

S1 -32B model is built in QWNE2.5-32B-INSTRUCCION, Alibaba’s Open-Bill of Model-Source-Source-Source-Source-soup-prower-soup-soup-soup-pro- It’s excellent in encryption, score 84.0 in Massurek, and in mathematics. Openitai Deepseek levertaked Chatgpt for the ordinary conduct of the terms of the Terms of Openai’s use is detected.

[Read More: DeepSeek’s 10x AI Efficiency: What’s the Real Story?]