Alibaba Challenges DeepSeek: New AI Qwen 2.5 launched
Alibaba launches Qwen 2.5, a new AI that outperforms DeepSeek. Competition between Chinese tech giants accelerates the evolution of artificial intelligence.
Ecommerce giant Alibaba has entered the world of AI by announcing the release of its AI called Qwen 2.5. claiming that its capabilities have left DeepSeek behind.
Gwen 2.5 is released on the first day of the lunar year, demonstrating the pressure DeepSeek has exerted not only on international rivals, but also on the domestic market.
Alibaba posted the news on its official WeChat account talking about the artificial intelligence models modernised by Meta and OpenAI. The news claimed that the IA Qwen outperformed almost everyone in the market, notably DeepSeek, ChatGPT-4o and Llama.
DeepSeek released the first version on 10 January, supported by the DeepSeek-V3 model. Subsequently, the company launched the R1 model, which attracted a lot of attention in the AI industry, causing the shares of technology companies such as Nvidia, RIOT and MARA to plummet. The low development and operational costs attracted investors, raising questions about the huge expenses of major AI companies in the US.
DeepSeek also shook up competitors, allowing them to upgrade their AI models. Two weeks after the release of R1, another competitor, TiTok's ByteDance, released an update to its AI model.
According to analysis of R1's performance on AIME, it outperformed the Microsoft-supported OpenAI project. AIME is a benchmark test that measures how effectively AI models understand and execute complex commands.
The DeepSeek V3 model's predecessor, V2, was released in May 2024. The release triggered a price war in China. DeepSeek V2 was an open-source project that cost about 1 Yuan per million tokens. This led Alibaba's team to announce price reductions of up to 97% in its model list.
Artificial Intelligence War in China
Other Chinese tech giants have also followed the AI trend, with companies such as Baidu releasing their own version of ChatGPT in 2023. Baidu also owns shares in Tencent, China's most valuable internet provider company.
Liang Wangfeng, founder of DeepSeek, said in an interview with a Chinese media outlet that the startup was not concerned about money. Its goal was to achieve AGI (General Artificial Intelligence).
AGI refers to autonomous systems that have the ability to outperform humans in tasks that are mundane, but nevertheless valuable.
It is worth noting that while large companies such as Alibaba Group have millions of employees, DeepSeek was founded by young engineers, scientists with PhDs and graduates from top Chinese universities.
In July, Wangfeng compared the high costs and hierarchical structures to DeepSeek's more flexible management practices, pointing out that these large companies are not suitable for AI.
The Pi Core Team released update 0.5.4, but the technical progress is in stark contrast to the on-chain reality. Low Mainnet activity and the sale of the WorkforcePool hackathon winner signal serious problems.
In a live trading challenge, DeepSeek Chat V3.1 outperformed all AI and Bitcoin models, growing by $35%$ in just three days. Learn about the winning strategy and failures of other competitors.
The ECB selected a Portuguese company specialising in artificial intelligence to develop fraud protection systems in an initiative to safeguard the central bank's future digital currency.
Ecommerce giant Alibaba has entered the world of AI by announcing the release of its AI called Qwen 2.5. claiming that its capabilities have left DeepSeek behind.
Gwen 2.5 is released on the first day of the lunar year, demonstrating the pressure DeepSeek has exerted not only on international rivals, but also on the domestic market.
Alibaba Claims Artificial Intelligence Outperforms DeepSeek
Alibaba posted the news on its official WeChat account talking about the artificial intelligence models modernised by Meta and OpenAI. The news claimed that the IA Qwen outperformed almost everyone in the market, notably DeepSeek, ChatGPT-4o and Llama.
DeepSeek released the first version on 10 January, supported by the DeepSeek-V3 model. Subsequently, the company launched the R1 model, which attracted a lot of attention in the AI industry, causing the shares of technology companies such as Nvidia, RIOT and MARA to plummet. The low development and operational costs attracted investors, raising questions about the huge expenses of major AI companies in the US.
DeepSeek also shook up competitors, allowing them to upgrade their AI models. Two weeks after the release of R1, another competitor, TiTok's ByteDance, released an update to its AI model.
According to analysis of R1's performance on AIME, it outperformed the Microsoft-supported OpenAI project. AIME is a benchmark test that measures how effectively AI models understand and execute complex commands.
The DeepSeek V3 model's predecessor, V2, was released in May 2024. The release triggered a price war in China. DeepSeek V2 was an open-source project that cost about 1 Yuan per million tokens. This led Alibaba's team to announce price reductions of up to 97% in its model list.
Artificial Intelligence War in China
Other Chinese tech giants have also followed the AI trend, with companies such as Baidu releasing their own version of ChatGPT in 2023. Baidu also owns shares in Tencent, China's most valuable internet provider company.
Liang Wangfeng, founder of DeepSeek, said in an interview with a Chinese media outlet that the startup was not concerned about money. Its goal was to achieve AGI (General Artificial Intelligence).
AGI refers to autonomous systems that have the ability to outperform humans in tasks that are mundane, but nevertheless valuable.
It is worth noting that while large companies such as Alibaba Group have millions of employees, DeepSeek was founded by young engineers, scientists with PhDs and graduates from top Chinese universities.
In July, Wangfeng compared the high costs and hierarchical structures to DeepSeek's more flexible management practices, pointing out that these large companies are not suitable for AI.
Sign up for Spaziocrypto®
No spam. Unsubscribe anytime.
Read Next
Pi Network between Technical Advancement and Mainnet Frustration: Price Drops 10%
The Pi Core Team released update 0.5.4, but the technical progress is in stark contrast to the on-chain reality. Low Mainnet activity and the sale of the WorkforcePool hackathon winner signal serious problems.
DeepSeek Chat V3.1 Dominates Crypto Market with 35% Return in Three Days
In a live trading challenge, DeepSeek Chat V3.1 outperformed all AI and Bitcoin models, growing by $35%$ in just three days. Learn about the winning strategy and failures of other competitors.
ECB chooses an AI start-up to prevent digital euro fraud
The ECB selected a Portuguese company specialising in artificial intelligence to develop fraud protection systems in an initiative to safeguard the central bank's future digital currency.
Google launches an open-source protocol for payments between AI agents
Google presents an open-source AI protocol for autonomous payments with stablecoins and cards.