Zhipu AI to Focus on AGI Research Over Short-Term Commercialization

Chinese AI company Zhipu AI is shifting its strategic focus towards advancing Artificial General Intelligence (AGI) research, foregoing short-term commercialization to pursue long-term technological breakthroughs. The company has launched a new initiative dubbed "Touch High" (or "Reach High") to spearhead these efforts.

According to an internal letter from founder Tang Jie, Zhipu AI plans to heavily invest in four key areas: long-horizon tasks, autonomous agent systems, fully self-training models, and extreme safety governance. A substantial portion of resources, reportedly in the billions of yuan, will be allocated to "mechanical explainability" to enhance the transparency of AI decision-making.

This strategy is characterized as "counter-intuitive" in the current market climate, which often favors rapid commercialization. Zhipu AI believes that pushing the boundaries of "intelligence's upper limit" is critical for future advancements in the AI field.

The company's roadmap for AGI development includes tackling challenges such as enabling AI to perform extended, complex tasks, creating fully autonomous intelligent agents capable of independent operation, and fostering systems that can train themselves. Zhipu AI also released its latest open-source model, GLM-5.2, highlighting its commitment to an open ecosystem and broader accessibility of AI.

By prioritizing fundamental AGI research and development, Zhipu AI aims to contribute significantly to the field while ensuring the safe and responsible evolution of artificial intelligence on a global scale.