Alibaba Group’s Hong Kong-listed shares surged by more than 8% on Thursday following the release of its new artificial intelligence (AI) reasoning model, QwQ-32B. The company claims that the model, with 32 billion parameters, delivers performance comparable to global AI hits like DeepSeek’s R1, which has 671 billion parameters.
The announcement was made through Alibaba’s AI unit on X, the platform formerly known as Twitter, where the company highlighted the QwQ-32B’s abilities in areas such as mathematical reasoning, coding, and general problem-solving. The model was put to the test in benchmark evaluations, performing on par with top AI models like OpenAI’s o1 mini and DeepSeek’s R1.
Alibaba’s new model is accessible via its chatbot service, Qwen Chat, where users can choose from a variety of Qwen models, including the powerful Qwen2.5-Max. The launch comes at a time when the Chinese government is increasing its support for industries, including artificial intelligence, humanoid robots, and 6G telecom.
DeepSeek, which has emerged as a key player in China’s AI landscape, continues to compete with global AI giants like OpenAI, offering models that rival the performance of more expensive alternatives with fewer computing resources.
In addition to Alibaba’s advancements, another AI release attracting attention was the introduction of Manus, an AI agent developed by the Chinese startup Monica. Manus, which outperformed OpenAI’s Deep Research in benchmarks for AI assistants, can help users with tasks such as travel planning and insurance comparisons. Currently by invitation only, a video showcasing Manus has gained significant interest, with over 280,000 views as of Thursday.