Chinese AI app DeepSeek’s entry into the artificial intelligence chatbot space has shifted the landscape, giving OpenAI’s ChatGPT a fierce challenge for dominance in innovation and resources.
DeepSeek surged into headlines after releasing its latest open-source model, DeepSeek-R1, which topped Apple App Store’s free downloads in both China and the United States on Monday, January 27, 2025. This milestone saw it outpace ChatGPT in free U.S. downloads—remarkable given ChatGPT’s home-based advantage.
The breakthrough behind DeepSeek-R1 lies in its use of pure deep learning methods to enable AI systems to develop reasoning capabilities spontaneously. These innovations, which OpenAI has controversially labeled “innovation theft,” have pressured the accuser to make some of its premium models freely available worldwide.
The world now witnesses not just competition but a sharp pivot from technological rivalry to an all-out AI war between the two global tech hubs.
China has a reputation: it excels at refining prototypes into products that often surpass their originals. But as chatbot users migrate en masse to DeepSeek’s Large Language Models (LLMs), can the company sustain this momentum and retain its users?
The Comparison
Image Generation: At the time of publication, ChatGPT still holds features DeepSeek lacks. For instance, DeepSeek initially couldn’t generate images—until Tuesday, January 28, 2025, when it defied analyst scepticism by launching Janus-Pro, which it described as a “revolutionary” image-generating tool.
The implications are staggering: the release of DeepSeek-R1 alone triggered a $500 billion collapse in U.S. tech stocks. If Janus-Pro delivers on its promises, the ripple effects could be huge.
Notably, while ChatGPT charges for image generation, Janus-Pro is unrestricted.
Prompt Performance: Tester reviews comparing ChatGPT-4o, o1, Claude 3.5 Sonnet, and DeepSeek reveal nuances. One noted ChatGPT’s edge in research-related tasks, followed by DeepSeek, with Claude trailing. Claude, however, shines in writing and explanatory tasks.
Coding: For coding, a Reddit user’s breakdown sums it up:
“DeepSeek R1 delivers exactly what I need in R or Python. o1 gets there but overcomplicates or skips steps. Claude requires more back-and-forth but writes robust code. 4o lags behind.”
Budget: DeepSeek’s frugality is revolutionary. Reports claim it spent just $6 million to build a generative AI model rivalling ChatGPT’s billion-dollar counterparts. This cost efficiency could redefine how LLMs are developed.
The Big Question
With its lean budget and rapid gains, can DeepSeek maintain its lead over ChatGPT?
One thing is certain: whichever tool you use, ensure it doesn’t outsmart you. Once that happens, discerning superiority becomes impossible.