Chinese tech giant Alibaba has unveiled its Qwen 2.5-Max AI model, boldly claiming it has surpassed DeepSeek’s well-regarded DeepSeek-V3 model. Announced on the first day of the Lunar New Year, the launch of Qwen 2.5 Max highlights the growing competition in China’s AI space and the pressure DeepSeek’s rapid rise has placed on both international and domestic competitors.
On WeChat, Alibaba’s cloud unit shared that the Qwen 2.5-Max outperformed OpenAI’s GPT-4, DeepSeek-V3, and Meta’s Llama-3.1-405B on several performance benchmarks. The timing of this release, during a period when most of China is celebrating the New Year, underscores the urgency Chinese companies feel to stay competitive with DeepSeek, which has caused waves in the AI market since its launch in early January.
DeepSeek’s impact on the market and competitors
DeepSeek’s sudden success, starting with the launch of its AI assistant powered by the DeepSeek-V3 model on January 10 and the R1 model on January 20, has shaken the tech world. The Chinese startup’s low-cost approach to developing powerful AI has sparked concern in Silicon Valley, especially as investors question the high development costs of leading US companies. In response, Chinese competitors are scrambling to enhance their models.
ByteDance, for instance, updated its flagship AI model shortly after DeepSeek’s R1 release, claiming it outperformed OpenAI’s GPT-1 in the AIME benchmark test, which evaluates AI’s ability to understand and respond to complex instructions. This move shows how DeepSeek’s rapid rise has spurred other domestic companies into action, with Alibaba’s latest release coming in response to the challenge posed by DeepSeek’s innovations.
A new competitor emerges: The Kimi k1.5 model from Moonshot.
Moonshot AI’s new model, Kimi k1.5, adds further complexity to the race, launched just days after DeepSeek’s R1. Kimi k1.5 is being hailed as a direct competitor to DeepSeek’s models and OpenAI’s GPT-4, with reports suggesting it outperforms both on key benchmarks. Unlike DeepSeek-R1, which lacks multimodal capabilities, Kimi k1.5 is a multimodal model that can process and reason across text, images, and code, giving it a significant edge in tasks requiring visual and textual data.
Kimi k1.5 has also been built at a fraction of the cost it would take to develop a similar frontier AI model in the US. This makes it an attractive alternative to other leading AI models, positioning Moonshot AI as a rising force in the global AI space. The model’s advanced reinforcement learning techniques enhance its capabilities, making it highly versatile for various applications.
Changing dynamics of China’s AI industry
China’s growing influence in AI is evident as companies like DeepSeek, Alibaba, and Moonshot AI are challenging the traditional dominance of US tech giants. The launch of DeepSeek-V2 last May triggered an AI price war in China, and Alibaba’s cloud unit responded by slashing prices by up to 97%. This price-cutting strategy has become commonplace among Chinese firms, including Baidu and Tencent, as they race to develop and release AI models that compete with OpenAI and other US-based giants.
DeepSeek, under the leadership of Liang Wenfeng, has taken a different approach, operating more like a research lab with a lean team of graduates and PhD students. Liang’s vision of achieving artificial general intelligence (AGI) with much lower overhead than larger tech companies contrasts with the high-cost, hierarchical models of Alibaba and other Chinese tech giants.
As China’s AI sector rapidly evolves, its influence on the global market continues to grow. The competition between DeepSeek, Moonshot AI, and Alibaba marks a significant shift in the AI landscape, with these startups and tech giants pushing the boundaries of what’s possible in AI development. The race for AI supremacy is on, and China is leading the charge.