The announcement coincided with the Chinese government’s renewed commitment to boosting AI and other emerging technologies, so as to enhance its competitiveness in the global tech race.

The QwQ-32B model, developed by Alibaba’s AI division, Qwen, is designed to rival some of the most powerful AI models available. Despite having 32 billion parameters, it reportedly delivers performance comparable to DeepSeek’s R1, a model with 671 billion parameters. 

Qwen stated via X that QwQ-32B is great in mathematical reasoning, coding, and general problem-solving, placing it among top AI models like OpenAI’s o1 mini and DeepSeek’s R1.

The model is accessible through Qwen Chat, Alibaba’s AI-powered chatbot service, which allows users to choose from multiple AI options, including the latest Qwen2.5-Max, the most advanced model in the Qwen lineup. 

In a move to encourage wider adoption and innovation, Alibaba has made QwQ-32B available as an open-source model on platforms such as Hugging Face and ModelScope under the Apache 2.0 license.

Government Backing and Market Response

Alibaba’s announcement followed a policy statement from the Chinese government pledging more support for industries such as AI, humanoid robots, and 6G telecommunications. 

This shows China’s strategy to build an AI-driven economy that extends beyond research and into real-world applications.

“China is rapidly building an application-driven AI ecosystem that isn’t just about research — it’s about immediate, tangible economic impact,” said Sun Wei, principal AI analyst at Counterpoint Research. 

Analysts believe these government-backed initiatives will accelerate AI adoption across industries, particularly in government agencies and smaller enterprises.

The market reacted positively to Alibaba’s latest AI development, with the company’s shares climbing to HK$140.5. The announcement also had a ripple effect on China’s tech stock index, which saw notable gains.

Alibaba’s AI Investment Strategy

Alibaba has committed to investing over 380 billion yuan ($52 billion) in AI infrastructure, including data centres, over the next three years. 

This investment shows its ambition to remain a leader in the field while competing with other players in China’s AI space.

Meanwhile, DeepSeek’s R1 model, which uses a sparse mixture-of-experts (MoE) transformer approach, continues to attract attention. In activating only a fraction of its 671 billion parameters per task, it provides a cost-effective alternative to models requiring more computing power.

According to the research, 74.2% of DeepSeek-R1’s outputs share stylistic fingerprints with OpenAI’s technology, raising talks about possible reliance on OpenAI’s model during training.

This revelation has also led to discussions around data sourcing, intellectual property rights, and transparency in AI development. If DeepSeek-R1 was trained using OpenAI-generated content without disclosure, it could cause legal and ethical risks, including reinforcing biases and limiting diversity in AI-generated text.

The study employed an advanced text attribution method, utilising three independent AI classifiers trained on outputs from OpenAI, Gemini, Claude, and Llama. To ensure accuracy, a classification was only confirmed when all three classifiers reached the same conclusion. This approach resulted in a 99.88% precision rate, with a false-positive rate of just 0.04%.

During testing, DeepSeek-R1’s texts were found to align with OpenAI’s writing style in 74.2% of cases. In contrast, Microsoft’s Phi-4 model exhibited a 99.3% disagreement rate with existing AI-generated texts, indicating independent training.

Shai Nisan, Copyleaks’ chief data scientist, commented on the importance of the findings, stating, “With this research, we have moved beyond general AI detection as we knew it and into model-specific attribution, a breakthrough that fundamentally changes how we approach AI content.”

The research team, led by Yehonatan Bitton, Shai Nisan, and Elad Bitton, adopted a rigorous “unanimous jury” approach to ensure reliability of their findings. Their method went beyond identifying known AI models to also detecting previously unseen ones by analysing unique stylistic markers.

If DeepSeek-R1’s model was developed using OpenAI’s work without proper attribution, it could mislead investors and stakeholders about the originality of its technology. 

This ultimately points to cautiousness about AI governance, competitive fairness, and the risks of intellectual property infringement in the industry. Transparency in model training and attribution is highly important in maintaining trust and ensuring ethical development practices.