Alibaba released a new AI model, claimed better performance than DeepSeek V3

Alibaba has released a new version of its Qwen 2.5 artificial intelligence model. The company claims that it performs better than almost all GPT-4o DeepSeek-V3. This comes after DeepSeek attracted attention for its AI model. DeepSeek has increased the competition not only abroad but also in its domestic market.

Wed, 29 Jan 2025 07:46 PM (IST)
 0
Alibaba released a new AI model, claimed better performance than DeepSeek V3
Alibaba released a new AI model, claimed better performance than DeepSeek V3

On Wednesday, Chinese technology firm Alibaba 9988.HK unveiled a new model of its Qwen 2. According to the news agency Reuters, the company has claimed its artificial intelligence model performs better than the popular DeepSeek-V3. The timing for 'Qwen 2.5-Max' is a little weird since today is the first day of the Lunar New Year, and at this time most Chinese citizens are on holiday and spend their time with friends and family. These weird times speak for themselves of pressure created by a sudden rise in Chinese AI startup DeepSeek, which, as you see, put pressure not only on foreigners but also on domestic competitors.

"Qwen 2.5-Max outperforms GPT-4o, DeepSeek-V3, and Llama-3.1-405B in almost all sectors," Alibaba's cloud unit said in an announcement posted on its official WeChat account, referring to the most advanced open-source AI models from OpenAI and Meta.

DeepSeek's AI assistant powered by the DeepSeek-V3 model was released on January 10, followed by the release of its R1 model on January 20. This has surprised Silicon Valley and led to a decline in tech stocks. The Chinese startup's reportedly low development and usage costs have led investors to question the heavy spending plans of leading AI firms in the United States.

DeepSeek's success abroad has also sparked a race among its domestic competitors to upgrade their own AI models.

Two days after the release of DeepSeek-R1, TikTok owner ByteDance released an update to its flagship AI model that it claimed performs better than Microsoft-backed OpenAI's o1 in AIME, a benchmark test that measures how well an AI model understands and responds to complex instructions.

This claim is similar to that of DeepSeek, where the company claimed its R1 model performed as well as OpenAI's o1 model in several tests that measure performance.

The earlier version of DeepSeek's V3 model, DeepSeek-V2, which was launched last May has thrown China into a price war. DeepSeek-V2 was open-source and cheap-it charged only 1 yuan ($0.14) for every 1 million tokens. Alibaba's cloud unit responded to this by announcing price cuts as high as 97 percent on a number of models.

Other Chinese tech companies also followed the same path, which include Baidu 9888.HK. The company provided the first chatGPT model for China in March 2023. Another prominent internet firm from China that has featured on this list is Tencent 0700.HK. Amongst the latest trending AI models raging in China, Baidu's Ernie Bot 4.0, ByteDance's Doubao 1.5 Pro, and Moonshot AI's Kimi k1.5 AI models can be mentioned.

Muskan Kumawat Journalist & Writer