The article by Mr. Dinh Hong Ky was published in the “Perspectives” section of VnExpress on January 30, 2025.
Based on my personal experience, I believe DeepSeek is still catching up with OpenAI and cannot yet be considered superior. Currently, DeepSeek R1 — their most powerful AI model — is roughly equivalent to OpenAI’s o1, while OpenAI has already developed up to OpenAI o3, which is close to achieving Artificial General Intelligence (AGI).
However, in recent days, DeepSeek has been frequently mentioned on tech forums and social media. Many blame it for financial market volatility, from Nasdaq drops to Bitcoin price falls. While some experts remain cautious in their assessments, many AI enthusiasts see DeepSeek as an emerging competitor capable of directly challenging OpenAI.
In my view, many perceive DeepSeek as smarter simply because most users only access GPT-4o for free, which has more limited reasoning capabilities. Since DeepSeek R1, comparable in power to OpenAI o1, is offered for free, users experience it as superior to OpenAI’s free version. If OpenAI o1 were also freely available, the debate might be less heated.
What’s more widely agreed upon is that DeepSeek is triggering a price revolution in the AI sector. While OpenAI operates on a traditional business model—offering premium, expensive versions—DeepSeek takes a completely different approach: slashing prices to the minimum.
Specifically, DeepSeek R1 matches the power of OpenAI o1 but costs only 1/30th of the price. Earlier, when DeepSeek V3 launched at just 1/11th the cost of GPT-4o, many considered it a shock. But R1 pushes the discount strategy to a whole new level.
Besides being cheap, DeepSeek offers APIs at prices far lower than OpenAI’s, causing many businesses to rethink their AI costs. They even allow free web access to DeepSeek R1, while OpenAI is still struggling with “limited free” policies. This pressure recently led OpenAI to announce partial free access to OpenAI o3 to stay competitive.
The AI race in China is heating up as Alibaba unexpectedly launched Qwen 2.5 during Lunar New Year, claiming it offers better reasoning, faster processing, and higher accuracy than DeepSeek V3, posing a major challenge to other open-source AI models.
This move shows Alibaba doesn’t want DeepSeek to monopolize the market. If DeepSeek is seen as a “new F1 racer” challenging OpenAI, Qwen 2.5 is like another “supercar” entering the race. The AI competition is no longer just between China and the U.S.; it’s also an internal battle among China’s tech giants.
DeepSeek demonstrates that being the biggest company doesn’t guarantee victory. While OpenAI leverages massive hardware and data to build powerful AI models, DeepSeek focuses on optimizing models for higher efficiency at lower cost. But with Alibaba’s entry, DeepSeek now faces significant pressure at home.
In the coming months, giants like Google, Meta, and Microsoft might well learn from, improve, and replicate the strategies of DeepSeek and Alibaba. Especially with DeepSeek and Alibaba opening their codebases, the AI community can quickly find similar optimizations.
The emergence of DeepSeek and Qwen 2.5 is changing perceptions of Chinese technology. Where many once saw Chinese tech firms primarily as Western copycats, names like ByteDance (TikTok), Huawei, DeepSeek, and Alibaba are now proving they can innovate independently.
Beyond impacting the AI race, rapid advances in models like DeepSeek, Qwen 2.5, and GPT-4o raise questions about the future of human jobs. In programming, AI models increasingly write code, fix bugs, and even optimize algorithms autonomously.
Repetitive jobs will inevitably be replaced by AI sooner or later. But programmers with problem-solving skills and strong business context will find AI to be a powerful assistant rather than a threat.
Similarly, AI will favor those who know how to use it, offering a huge competitive advantage. In the future, programmers may shift from just coding to software architects or problem-solving specialists, with AI as their powerful tool.
The race among OpenAI, DeepSeek, and Alibaba has only just begun. While DeepSeek attracts attention with its low price and open-source model, OpenAI still holds advantages in resources and technology, and Alibaba is showing its strength with Qwen 2.5.
Going forward, AI will become cheaper, stronger, and more widespread. Users will have more choices, costs will drop significantly, and those who leverage AI effectively will be the winners.
One thing is certain: the AI arena is no longer America’s exclusive stage — China has officially entered with an ambitious strategy.
