Is ChatGPT losing its edge? Chinese chatbots challenge US tech dominance

Alibaba’s recent decision to slash the cost of its Tongyi Qianwen chatbot—commonly known as Qwen and categorised as a large language model (LLM)—by a staggering 85 percent is more than just a pricing move.

For all practical purposes, it’s a smart gambit to dominate the next phase of the US-China tech rivalry.

While global attention has largely been on American giants like OpenAI, Alphabet’s Google, and Meta, China’s tech leaders—including Baidu, Tencent, and Huawei—have been steadily advancing their own AI technologies to challenge the long-standing dominance of US companies.

Alibaba’s price reduction for its LLM services comes as Chinese technology companies strive to make artificial intelligence more accessible to businesses and developers.

In a country with a population of over one billion, this step may add a new layer to China’s growing AI sector, where domestic firms are creating systems tailored to the world’s largest internet market.

Can China’s rapid progress with cheaper subscription options put overpriced American LLM companies behind?

Chinese Large Language Models (LLMs)

Large language models are a type of artificial intelligence trained on vast amounts of text data and have become the sophisticated tools used across various aspects of our daily life.

One of the most recognisable applications of LLMs in everyday life is ChatGPT, the world’s most used generative AI chatbot with 200 million users weekly.

But does being the most popular necessarily mean being the most capable?

One might think that the ban on ChatGPT and Gemini in China would put the world’s second-largest economy behind in the race, but it has actually achieved the opposite result with its new models.

Despite regulatory restrictions, including the bans, the Cyberspace Administration’s approval of over hundred domestic AI models by early 2024 highlights the country’s thriving AI market.

Chinese LLMs began making strides in the industry in 2023, starting with Baidu’s Ernie Bot. Initially met with heavy criticism, Baidu persisted in refining and advancing the bot.

Baidu’s Ernie Bot leads the pack with 300 million users – twice Germany's population – offering unparalleled performance in understanding Chinese idioms, dialects, and cultural nuances.

Its precision is exemplified by correcting ChatGPT’s error about the hometown of Liu Cixin, author of The Three-Body Problem, adapted into a popular Netflix series.

Baidu’s latest iteration, Ernie 4.0, is touted as rivalling GPT-4 and proves significant progress in understanding and reasoning.

“Baidu is an AI company with a strong internet foundation,” Beijing-based Baidu boss Li told TIME in August.

“So we have lots of users [and] scenarios … to improve our foundation model at a much faster pace.”

Meanwhile, Alibaba’s Tongyi Qianwen has developed a variety of versions that can carry out different tasks. Some of its versions provide open-source and enterprise-focused versions, enabling businesses to handle tasks such as content creation, solving mathematical problems, and even permitting some to be openly downloaded.

However, the Chinese LLM industry is not just about Baidu’s Ernie or Alibaba’s Qwen. China’s advancements in AI have given rise to a host of innovative large language models (LLMs), each created to meet diverse needs across industries.

The sector has paved the road for many chatbots, such as Tencent’s Hunyuan, Huawei’s Pangyu, and TikTok app owner ByteDance’s Doubao.

The world's largest video game vendor and the owner of the biggest Chinese messaging app WeChat, Tencent integrates Hunyuan deeply with its ecosystem, particularly within WeChat, by making this feature readily accessible to its everyday users.

Tencent has said Hunyuan has strong Chinese language processing abilities and “advanced” logical reasoning.

Tech conglomerate Huawei took a slightly different approach to its rivals and built Pangu AI models to specialise in industry-specific applications including government, finance, manufacturing, mining, and meteorology.

In one instance, Huawei said its Pangu Meteorology Model can predict the trajectory of a typhoon over 10 days in around 10 seconds, rather than the four-to-five hours it took previously.

The owner of Tiktok, one of the world's most popular social media platforms, has built an AI model called Doubao at a price much cheaper than other companies. The Doubao model has the capacity to generate voices as well as generate code for developers, among other capabilities.

“China's top domestic AI large models have reached the level of GPT-3.5, and the technical gap with GPT-4 is narrowing,” said Dou Dejing, an adjunct professor at Tsinghua University's Electronic Engineering Department.

The pricing gambit

Alibaba’s drastic reduction in the cost of its Tongyi Qianwen model reflects a commitment to widespread AI adoption.

This move is part of a broader trend among Chinese companies to make advanced technology accessible to a larger audience. Baidu’s Ernie Bot is available through its cloud services, enabling small businesses to integrate AI into their operations at reduced cost.

In contrast, US firms like OpenAI rely on subscription-based models, which cater to premium markets but limit accessibility for smaller players.

By focusing on affordability, Chinese companies are well-positioned to capture market share in emerging economies and among cost-sensitive businesses globally.

However, sustaining these low prices will require balancing operational costs with market expansion.