China’s AI Models Are 85% Cheaper Than U.S. Rivals — Why It Matters

On the last day of 2024, and for the third time in a year, Alibaba Cloud slashed the prices of generative AI products by a staggering 85%.

Alibaba Cloud, a division of Alibaba Group, announced the new prices for its text and visual multimodal AI model Qwen-VL, with a rapidly growing domestic artificial intelligence (AI) industry cited as the reason.

Chinese regulators have approved more than 252 GenAI services for public use.

From Huawei to ByteDance (TikTok developer), Tencent, Baidu, and other local startups, large language models (LLMs) and advanced multimodal models have flooded the national Chinese market in the past 18 months.

Alibaba’s multimodal AI Qwen-VL now costs $0.41 per million input tokens, a price set to match ByteDance’s recently released model — also at $0.41 per million input tokens.

Compared to OpenAI’s GPT-4.o, priced at $2.50 per million input tokens, Alibaba Cloud’s new multimodal AI is 83.6% cheaper despite offering similar performance metrics.

In this report, we compare AI models made in China against those developed in other countries, especially the U.S., to understand the state of global AI development.

Table of Contents Table of Contents

1. Key Takeaways
2. Affordable AI Solutions in China Disrupting the Industry
3. The Impact of Chinese AI on Global Markets
4. The Trade-Off Factors: Are China AI Models Biased and Censored?
5. U.S. Tech Companies Cooperation with China’s AI Industry
6. Should U.S. AI Developers and Governments Be Worried?
7. The Bottom Line
8. FAQs
9. References

Show Full Guide

Table of Contents

1. Key Takeaways
2. Affordable AI Solutions in China Disrupting the Industry
3. The Impact of Chinese AI on Global Markets
Show Full Guide
4. The Trade-Off Factors: Are China AI Models Biased and Censored?
5. U.S. Tech Companies Cooperation with China’s AI Industry
6. Should U.S. AI Developers and Governments Be Worried?
7. The Bottom Line
8. FAQs
9. References

Affordable AI Solutions in China Disrupting the Industry

Besides Alibaba’s Qwen family of AI models, which exhibit powerful context understanding and multilingual support, there are several other noteworthy companies that have carved their unique path to powerful AI models in China.

Baidu — the Chinese search engine giant — recently renamed its flagship AI bot. With its Wen Xin Yi Yan AI, Baidu uses its market share to position itself as the major player in the field of AI web search assistants.

This type of market is considered the future of web browsing, with companies like OpenAI, Google, and Apple moving in this direction as well.

Another major tech company to consider in China is Tencent. Tencent’s Hunyuan is a large-scale multimodal model. The AI supports text-to-image generation and excels in logical reasoning, chat, and content creation. Their approach and business model are similar to those of OpenAI. With a wide-ranged portfolio, Tencent seeks to be a dominant force in the Chinese AI market.

Interestingly, China’s local AI startups are also breaking ground beyond all expectations. The upcoming and increasingly popular startup Zhipu has been making waves with its different models, including its General Language Model (GLM) series. Zhipu AI also supports English and Chinese tasks and is growing in the domestic market.

Similarly, DeepSeek, another Chinese startup that has gained international recognition due to its demonstrated performance capabilities, is developing models at incredible speeds with extremely low budgets.

DeepSeek’s price for its API chat is $0.014 per 1M input tokens, significantly lower than Alibaba Cloud’s AI price of $0.41 per 1M input tokens — and let’s remind ourselves again that OpenAI charges $2.50 per 1M input tokens.

More importantly, DeepSeek, like many other Chinese AI companies, has jumped into the open-source AI movement, often offering their models through Hugging Face and GitHub for developers.

The Impact of Chinese AI on Global Markets

U.S. foreign policy has repeatedly attempted to curb the development of AI in China by restricting chip exports and engaging in sanctions. As the new Trump administration prepares to take office, new aggressive tariffs and restrictions are expected to emerge.

Despite these efforts, and contrary to what many may think, the AI industry in China is not struggling at all. In fact, it is thriving.

Proof of this is the domestically flooded market and the offering of products with high-performance capabilities for lower prices. As prices reach their bottom, AI companies in China will innovate to add advanced features to stay ahead of their competitors.

To understand the reasons behind the China AI boom, Techopedia spoke to Dev Nag, former CTO of Wavefront (acquired by VMware), former Senior Engineer at Google, and currently CEO and Founder of QueryPal, a support automation company.

Nag said:

“What’s fascinating about China’s AI sector is the three-pronged strategy that’s really setting them apart.”

Nag explained that Chinese AI companies are aggressively slashing prices, taking a much more relaxed approach to training data thanks to less restrictive IP laws, and implementing a regulatory framework that enables rapid deployment while maintaining state oversight.

“What’s especially interesting is their stronger commitment to open-source development, with companies like DeepSeek and Baidu being notably more transparent with their research than their Western counterparts.”

Nag said that the combination of these factors creates a remarkably fast innovation cycle within China’s AI ecosystem.

The Trade-Off Factors: Are China AI Models Biased and Censored?

Some argue that China-made AI models are trained on bias and ‘censored-by-government’ databases. Additionally, Chinese tech companies have a legal obligation to work with the government if requested.

Nag from QueryPal recognized that content restrictions and censorship are important trade-offs for developers to consider but added that from a purely technical standpoint, the achievements of China’s AI are impressive.

“Chinese AI models are showing some truly remarkable results. They’re not just matching Western counterparts, but are often outperforming them at a fraction of the cost.”

DeepSeek’s V3 model, for instance, is achieving comparable or superior results in specialized tasks like code generation and reasoning on the LLM Arena leaderboards while costing 95% less — and comparing well with open source code and open model weights vs. proprietary models like OpenAI’s o1-mini and Google’s Gemini 1.5 Pro.

U.S. Tech Companies Cooperation with China’s AI Industry

The commercial relationships between China and the U.S. are extremely complex. In this global scenario, not everything is black and white. Companies like Apple, which has a massive market share in Asia, often find creative ways to reach their Chinese customers.

On December 15, Reuters reported that Apple was in early exploratory talks with Tencent and ByteDance about integrating their Chinese-made AI models into iPhone models sold in the country.

While Apple in most parts of the world runs its flagship AI, Apple Intelligence, integrated with ChatGPT, in China, OpenAI models have not been approved by the government and are not allowed.

This regulatory landscape forces companies like Apple to seek local partners if they want to remain competitive in one of the biggest markets of the world.

Nag from QueryPal told Techopedia that Apple’s potential collaboration with Chinese AI firms is actually a fascinating harbinger of what’s likely to come.

“Despite all the geopolitical tensions, we’re probably going to see more U.S. companies pragmatically engaging with Chinese AI developers.

“The driving forces here are pretty compelling: there’s the sheer size of China’s market, the advanced capabilities Chinese firms have developed in specific domains, and the significant cost advantages they offer,” Nag said.

“Of course, these collaborations will need to carefully navigate U.S. export controls and Chinese data regulations.”

Should U.S. AI Developers and Governments Be Worried?

Is the significant volume of affordable AI solutions in China’s market a risk or challenge to global and U.S. industries?

According to Nag from QueryPal, yes, there are reasons to be concerned.

“What we’re seeing is unprecedented — these dramatic price reductions and rapid innovation cycles are putting serious pressure on the premium pricing models that Western companies have relied on.”

Nag spoke of a real risk of market share erosion for U.S. companies, particularly in developing markets, compounded by Chinese companies’ ability to operate profitably with much thinner margins while maintaining technological competitiveness across a variety of benchmarks.

“The U.S. strategy of constraining Chinese AI development through export controls and restrictions seems to have led to some unintended consequences, at least so far,” Nag said.

“Rather than slowing development, these restrictions appear to have accelerated China’s push for domestic innovation and self-sufficiency and increased state support for GenAI.”

The Bottom Line

While not everyone will agree that China is playing a leading role in shaping the next generation of AI models, the Western perception of the country’s AI industry playing catch up is way off the mark.

Chinese big tech and startups have capitalized on the AI movement and are rapidly iterating, innovating, and competing domestically.

The slashing of prices is not representative of the quality of Chinese AI models but of the maturity of its sectors

Turning to open source and making models available for anyone to use while demonstrating high-performance metrics, China’s AI industry is booming.

FAQs