Companies

Alibaba's Qwen Model: A Cost-Effective Alternative to DeepSeek

Published February 11, 2025

The Chinese tech giant, Alibaba Group Holdings Ltd, has made headlines with its Qwen model, which presents a more affordable option compared to DeepSeek. Recent developments from U.S. computer scientists have demonstrated that they successfully trained a new reasoning model for under $50. This achievement was greatly aided by Alibaba's open-source technology.

Researchers from esteemed institutions like Stanford University and the University of California, Berkeley, have tapped into Alibaba's Qwen 2.5-32b-Instruct model. Their goal? To reduce the financial barriers associated with AI training, as highlighted in a recent research paper.

The advancements of Alibaba's AI model showcase China's rapidly advancing technological capabilities within the ever-evolving AI landscape. This development comes closely after the unexpected launch of the R1 model by the Chinese AI startup, DeepSeek, which has gained attention for its high-performance and budget-friendly open-source AI solutions.

In September, Alibaba unveiled its open-source Qwen 2.5 series models, which feature a range of parameters from 0.5 billion to an impressive 72 billion. These models have been enhanced to improve performance, particularly in areas such as mathematics and coding, and they support more than 29 languages.

Overall, Alibaba's efforts not only help push the boundaries of what is possible with AI but also lower the costs associated with acquiring such technology. This is a significant step forward for researchers and developers alike, providing them with powerful tools to innovate and expand their projects without the heavy financial burden usually associated with advanced AI systems.

Alibaba, Qwen, DeepSeek