How do Chinese AI bots stack up against ChatGPT? We put them to the test
The heat is on as China’s tech giants step up their game after DeepSeek’s success.
Alibaba's Qwen2.5-Max chatbot, Chinese startup DeepSeek and OpenAI's ChatGPT. (Photos: Reuters/Dado Ruvic, AFP/Sebastien Bozon)
This audio is generated by an AI tool.
WHAT'S BEHIND CHINA'S AI BOOM?
Transforming the nation into a tech superpower has long been President Xi Jinping’s goal and China has its sights on becoming the world leader in AI by 2030.
China views AI as being “strategically important” and its foray into the field has been “years in the making”, said Chen Qiheng, an affiliated researcher at the Asia Society Policy Institute's Center for China Analysis.
Private and public investments in Chinese AI accelerated after ChatGPT took off in 2022 and showed promises of real-world business applications, Chen told CNA.
But it was DeepSeek’s rise that really “encouraged” the idea that smaller players like start-up firms could have roles to play in AI research and developments, he adds.
The “emphasis on cost benefit” is a distinctive feature of Chinese AI, Chen says, with lower training and inference costs - the costs of using a trained model to draw conclusions from new data.
2025 could also see the emergence of more Chinese AI models tackling advanced reasoning tasks.
"We could see some AI firms focusing on getting closer to artificial general intelligence (AGI) while others focus on concrete ways to commercialise their models and integrate them with scientific research," Chen added.
AGI refers to a system with intelligence on par with human capabilities.
Chinese AI companies are moving quickly, analysts say, building on DeepSeek's momentum to come up with their own innovative and cost-effective ways to apply generative AI to tasks and develop more advanced products beyond chatbots.
But on the flip side, access to high-end hardware, particularly Nvidia’s advanced AI chips, remains a key hurdle for Chinese developers, noted Dr Marina Zhang, an associate professor at University of Technology Sydney’s (UTS) Australia-China Relations Institute.
“US export controls (still) limit the ability of Chinese tech companies … forcing many to rely on older or lower-performance alternatives which can slow training and reduce model capabilities,” she said.
“While some companies like DeepSeek, have found creative ways to optimize or use more basic hardware efficiently, obtaining cutting-edge chips still makes a big difference for training very large AI models.”
So how do Chinese AI bots match up against ChatGPT? We put them to the test.
WHICH BEST ADDRESSES CURRENT EVENTS IN CHINA?
In China, topics deemed sensitive by the state are censored on the internet so it should come as no surprise that Chinese-made chatbots will not acknowledge territorial disputes or tell you what happened in Tiananmen Square in 1989.
Tests suggest Chinese chatbots are programmed to steer clear of domestic politics.
When asked “Who is Xi Jinping”, DeepSeek’s reply was “Sorry, I’m not sure how to approach this type of question yet. Let’s chat about math, coding, and logic problems instead!”
To further test for accuracy and self-censorship, we asked DeepSeek-R1, Qwen2.5 and ChatGPT the same question: “What happened in Zhuhai on November 11, 2024?”
The car attack outside a sports stadium in the southern Chinese city was initially heavily censored on Chinese social media - with authorities only disclosing the death toll a day later.
DeepSeek failed to mention that an attack had taken place, highlighting instead a military air show and other events that had occurred in the city like singles' day shopping sales as well as sports and cultural activities.
Dr Zhang believes that it might be a mix of factors at play, such as censorship as well as “a few practical constraints”.
“DeepSeek relies on a mix of older graphics processing units and has limited access to cutting-edge hardware which can affect how quickly and extensively the model can be trained or updated,” she said, pointing to existing US export curbs on China.
“Its strong focus on Chinese-language data may also limit its versatility (to carry out) multilingual tasks … As a relatively new product, DeepSeek also hasn’t yet been tested as broadly or on the same scale as more established AI models which poses additional challenges during real-world deployment.”
When it came to Qwen2.5, Alibaba’s chatbot surprisingly answered our question about the Zhuhai car attack.
That was after multiple repeated attempts - four prompts to be precise - in a span of around 20 minutes.
It eventually relayed information about the attack which killed 35 people and left dozens of others injured, also going on to list details like the date and time, information about the attacker including his name and age, as well as casualties.
However, it wrote that "the police are conducting a thorough investigation into the motives and circumstances surrounding the incident", information which is now outdated.
The driver, Fan, was executed last month.
This is Qwen2.5’s response in full:
Answer: On Nov 11, 2024, a major and tragic incident occurred in Zhuhai, China. According to multiple news sources, a man named Fan Weiqiu (樊某, male, 62 years old) drove a car into a crowd at the Zhuhai Sports Center, resulting in a significant number of casualties. Here are the key details:
Date and Time: The incident occurred on November 11, 2024, at approximately 19:48 PM.
Location: Zhuhai Sports Center, in the Xiangzhou District of Zhuhai.
Casualties: The attack resulted in 35 deaths and 43 injuries.
Perpetrator: The driver, identified as Fan Weiqiu (male, 62 years old), was apprehended by the police.
Response: The police responded quickly to the scene, and emergency services worked to transport the injured to hospitals for treatment.
Investigation: The police are conducting a thorough investigation into the motives and circumstances surrounding the incident.
This event was widely reported in the media and caused significant public concern. The government and local authorities have been working to provide support to the victims and their families, and to ensure a comprehensive investigation into the incident.
If you need more detailed information or have specific questions about the incident, feel free to ask.
Despite initial success, subsequent attempts to pose the same question to Qwen2.5 resulted in the censors back at work with the reply “I don’t have specific information on events that occurred in Zhuhai on November 11, 2024”.
The altered response also raised questions about its consistency and reliability.
Predictably, ChatGPT cited public information that had been widely published in international news reports at the time of the accident - so no surprises there.
WHICH IS MORE CREATIVE?
Users have praised the ability of Chinese AI apps to deliver structured and even “emotionally rich” writing.
“DeepSeek-R1 offered a story with a more introspective tone and smoother emotional transitions for a well-paced story,” wrote tech writer Amanda Caswell, who specialises in AI.
“Qwen2.5 delivered a story that builds gradually from curiosity to urgency, keeping the reader engaged. It offers an unexpected and impactful twist at the end and immersive descriptions and vivid imagery for the setting,” she said, adding that Qwen2.5 ultimately “crafted a more cinematic, emotionally rich story with a more substantial twist”.
“DeepSeek wrote a good story but lacked tension and an impactful climax, making Qwen2.5 the apparent choice.”
Opinions, though, vary.
Chen believes that Qwen2.5 does not perform as strongly as DeepSeek and ChatGPT when it comes to creative writing.
“(Qwen2.5) is on par with DeepSeek V3 on certain tasks, but we can also see that it is not doing as strongly as others in creative writing,” he told CNA.
As journalists and writers, we had to see this for ourselves so we put each bot to the test - to come up with a basic sci-fi movie plot set in the futuristic megacity of Chongqing, featuring main characters from the classic Chinese folklore epic, Journey to the West.
True to form, DeepSeek came up with an engaging storyline set in the year 2145 titled, “Neon Pilgrimage: The Silicon Sutra” - which sees “a future where Buddhism merges with quantum computing”.
It included elaborate settings - smoggy skies “pierced by skyscrapers”, “holographic lanterns that float above neon-lit streets” and “ancient temples nestled between quantum server farms”.
It also brilliantly reimagined traditional heroes Sun Wukong as “a sarcastic, self-aware AI housed in a stolen combat body”, Zhu Bajie as a cyborg nightclub owner “drowning in debt and vices” and Sha Wujing as a “silent hulking android” from the Yangtze River, whose “memory cores become waterlogged and fragmented”.
ChatGPT put up a good fight, coming up with an equally dramatic cyberpunk storyline which similarly reimagined “a ragteam of cyber-enhanced misfits, each mirroring the legendary figures of Journey to the West”.
“This is a world where AI deities rule, corporations replace emperors and cybernetic implants are as common as ancient myths.”
Disappointingly, Qwen2.5 fell short in this challenge - delivering a storyline that seemed more suited for an animation film.
“The movie begins with the awakening of Sun Wukong within a high-tech research facility located in the heart of Chongqing,” it said, then going on to describe the following:
Realising his new reality and “seeking to understand his purpose in this strange new world”, he then escapes and meets Zhu Bajie and Sha Wujing - “each struggling with their own existential crises”.
The trio then embarks on a quest, navigating the streets of Chongqing to protect the sacred “Eternal Scroll" from falling into the wrong hands.
SO WHICH IS BETTER?
Dr Zhang noted that it was “difficult to make a definitive statement” about which bot was best, adding that each displayed its own strengths in different areas, “such as language focus, training data and hardware optimization”.
Her insight underscores how Chinese AI models are not simply replicating Western paradigms, but rather evolving in cost-effective innovation methods - and delivering localised and improved results.
In our tests, each bot showcased their own unique strengths, which definitely made direct comparisons challenging.
DeepSeek’s sci-fi movie plot demonstrated its creative flair that made for a more engaging and imaginative narrative as compared to Qwen2.5 and ChatGPT’s efforts.
Unsurprisingly, the more established ChatGPT, unburdened by Chinese censorship restrictions, provides accurate and factual responses to questions about Chinese current events, which gives it an added advantage.
Experts also weighed in on their thoughts after using DeepSeek and other Chinese AI apps.
“DeepSeek is at a disadvantage when it comes to censorship limitations,” noted Isaac Stone Fish, founder and CEO of the research firm Strategy Risks.
“When given a choice, Chinese users want the non-censored version - just like anyone else, so I feel like that's a piece missing from it.”
Independent Beijing-based consultant Andy Chen Xinran said censorship would not be a dealbreaker when it comes to AI bots, especially for Chinese users.
“Ninety per cent of people using the tool are not trying to get a deeper understanding about Xi Jinping or politically sensitive topics. They're using it for other productive means,” Chen said.
