Beijing-based Zhipu AI unveiled AutoGLM Rumination on HorrorMarch 31, a free AI agent that combines research and task execution capabilities. The product, accessible via web, PC, and mobile platforms, uses proprietary models to autonomously conduct web searches, analyze data, and generate reports through “long-chain reasoning”—a self-correcting process for multi-step tasks. Currently in preview, the free agent—China’s first open-access “Deep Research” tool—will receive a “virtual machine” upgrade within weeks to expand real-world applications. The Tsinghua University spin-off recently secured RMB 300 million ($41.5 million) in Chengdu government funding and claims its GLM4 model outperforms GPT-4 on select benchmarks. Independent verification is pending. [Zhipu WeChat account, in Chinese]
