Alibaba unveils new AI model, claims superiority over DeepSeek, ChatGPT


Alibaba launched a new version of its artificial intelligence (AI) model, Qwen 2.5, claiming it outperforms one of the latest versions of ChatGPT and DeepSeek-V3, which made waves in global markets this week.

 

In a statement on its WeChat account, the Chinese tech giant said its Qwen 2.5-Max model surpasses OpenAI’s ChatGPT-4o, DeepSeek’s DeepSeek-V3, and Meta’s Llama 3.1-405B across almost all benchmarks.

 

Alibaba’s release of the model on the first day of the Lunar New Year holiday highlights the significant impact of DeepSeek’s latest AI model, which debuted this month, even on Chinese firms, according to Reuters.

 

On Jan. 10, DeepSeek launched an AI-powered personal assistant based on its DeepSeek-V3 model, followed by the release of DeepSeek-R1 on Jan. 20. Both models outperformed even the most advanced versions of ChatGPT. They were developed at a fraction of the cost compared to US counterparts.

 

This raised investor concerns over the efficiency of capital spending by American firms on AI development, given the vast amounts of funding pouring into the sector, as well as China's growing technological capabilities despite Western restrictions.

 

In addition to DeepSeek’s efforts, ByteDance, the owner of TikTok, also rolled out an update to its leading AI model on Jan. 22, claiming it surpassed ChatGPT-01 in tests measuring its ability to comprehend and respond to complex commands.

Comments 0

Sorry: the validity period has ended to comment on this news
Opinions expressed in the comments section do not reflect the views of Argaam. Abusive comments of any kind will be removed. Political or religious commentary will not be tolerated.
Call Request

Argaam Investment Company has updated the Privacy Policy of its services and digital platforms. Know more about our Privacy Policy here.

Argaam uses cookies to personalize content, to provide social media features and analyze traffic, that we might also share with third parties. You consent to our cookies if you use this website