Latest UniPat AI news, funding rounds and robotics & AI coverage. $300M raised. 2 articles tracked by The Robotics Media.
UniPat AI (legal name Pattern Star River (Beijing) Technology Co., Ltd.) builds evaluation benchmarks, synthetic training data and reinforcement-learning environments used to test and train large AI models in realistic, real-world scenarios rather than static lab tests. It was founded in Beijing in late 2025 by Li Kuan, a former researcher at Alibaba's Tongyi AI Lab, together with colleagues who previously worked at Alibaba, Moonshot AI and Tencent.
The company sells its benchmarking and data services to major Chinese AI labs, including Alibaba, Tencent, ByteDance, Moonshot AI and DeepSeek. Its benchmark suite covers coding agents, browser/computer-use agents, biomedical and scientific reasoning, and multimodal visual reasoning. In 2026 it also released EchoZ-1.0, a forecasting model trained with a "Train-on-Future" method and tested on real-money prediction markets such as Polymarket.
In September 2026, Alibaba agreed to lead a $300 million financing round in UniPat AI, joined by Tencent and existing investor HSG (formerly Sequoia Capital China), valuing the company at about $2.5 billion.