Alibaba unveils trillion-parametre AI model

Image Source: Unsplash

Chinese e-commerce giant Alibaba on Wednesday unveiled its most powerful artificial intelligence language model to date, underscoring its ambitions to challenge US dominance in advanced AI development.

The model, called Qwen3-Max, contains more than 1 trillion parametres — the variables that determine how an AI system processes information — and is designed to excel in code generation and autonomous agent capabilities, according to Zhou Jingren, chief technology officer at Alibaba Cloud. 

Autonomous agents are systems that can act independently toward user-defined goals, requiring fewer human prompts than traditional chatbots.

Alibaba said Qwen3-Max outperformed competitors including Anthropic's Claude and Chinese rival DeepSeek's V3.1 model in certain benchmarks such as Tau2-Bench.

The announcement was made at Alibaba's annual conference, where CEO Eddie Wu stressed that the speed of AI development had "far exceeded expectations" and signaled the company would boost investment in the sector. 

Earlier this year, Alibaba pledged 380 billion yuan (USD 53.4 billion) over three years to expand AI-related infrastructure, including cloud computing and data centres.

The company also introduced Qwen3-Omni, a multimodal system capable of powering immersive applications like smart glasses and intelligent cockpits.

Alibaba, best known for its sprawling e-commerce operations spanning marketplaces like Taobao and Tmall, has increasingly put AI at the centre of its strategy as it competes with Chinese peers such as Baidu, Tencent and DeepSeek, as well as US leaders OpenAI and Google DeepMind.