Alibaba Releases New Qwen Model, Says It Beats DeepSeek V-3

Alibaba Releases New Qwen 2.5 AI Model, Claims It Outperforms DeepSeek V-3

Disclosure

Updated: 29 January 2025

Don’t invest unless prepared to lose all the money you invest. This is a high-risk investment, you shouldn’t expect to be protected if something goes wrong.

Join Our Telegram channel to stay up to date on breaking news coverage

Chinese tech giant Alibaba has unveiled an upgraded Qwen 2.5 AI model, claiming it outperforms DeepSeek—the rival AI that sent Nvidia stock tumbling earlier this week.

In a Jan. 28 X post, the Alibaba cloud team said on the AI’s official X page that the new Qwen model “achieves competitive performance against the top-tier models, and outcompetes DeepSeek V3 in benchmarks like Arena Hard, LiveBench, LiveCodeBench, GPQA-Diamond.“

Rivals Reel From DeepSeek Revelations

Qwen 2.5-Max’s debut was unusual, coming on the first day of the Lunar New Year at a time when Chinese are off work. It underscores how the sensational claims made by DeepSeek are pressuring both foreign rivals like OpenAI and Meta, as well as local competitors.

Revelations about the performance of DeepSeek sent tech stocks crashing on Monday, with leading AI chip maker Nvidia losing almost $600 billion of its market cap in a single day, the most in stock market history.

Two days after the release of DeepSeek-R1 on Jan. 20, TikTok owner ByteDance released an update for its own AI model. The company then claimed its model outperforms OpenAI’s o1 model in AIME, which is a benchmark test used to measure how well AI models can comprehend and respond to complex instructions.

DeepSeek Does Not Care About Pricing, Just Wants To Achieve AGI

DeepSeek’s R1 launch hammered technology stocks after the company revealed that it cost less than $6 million to train its model that also runs on computer chips that are much cheaper than those used by US competitors OpenAI, Meta and X.

Despite the major alleged competitive advantage, DeepSeek’s founder said in an interview with a Chinese media outlet called Wave in July last year that the startup “did not care” about price wars. Instead, the company’s top priority is to achieve AGI (artificial general intelligence).

deepseek's r1 is an impressive model, particularly around what they're able to deliver for the price.

we will obviously deliver much better models and also it's legit invigorating to have a new competitor! we will pull up some releases.

— Sam Altman (@sama) January 28, 2025

OpenAI CEO Sam Altman commented on the R1 model in a Jan. 28 X post, and called it “impressive.” He added that OpenAI will soon “deliver much better models.”

Following the post, Microsoft, one of OpenAI’s largest investors, and OpenAI accused DeepSeek of using ChatGPT to train its R1 model, and has subsequently launched an investigation into the matter.