Penglai Zhishu

Hybrid Key-Value Database HetuKV

The Penglai Intelligent Digital R&D team, guided by the "software-defined storage" philosophy, independently developed HetuKV, a key-value database with full controllability. It maximizes flash memory performance and overcomes the bottlenecks of solutions heavily dependent on main memory. Fully compatible with all Redis data types, HetuKV serves as one of the domestic alternatives to foreign products like Redis.

Delivers an efficient, stable, and low-migration-cost alternative while offering unique advantages including higher storage node density, reduced energy consumption, lower total cost of ownership, decreased latency, and support for highly customized solutions.

HetuVDB Vector Database

As an AI company, we recognized the urgent need for a reliable, easy-to-deploy vector database to support large model training and applications. To address this, our team independently developed HetuVDB—a secure, reliable, and high-performance vector database.

Designed for multimodal vector data needs of large models, this product supports retrieval and storage of both dense and sparse vectors. It enables GPU acceleration and is compatible with mainstream machine learning frameworks like PyTorch and TensorFlow. Hecun Vector Database significantly improves the efficiency of storing and retrieving unstructured vector data for AI model inference and training, reduces access latency, and delivers a secure, reliable, high-performance storage solution for large-scale multimodal vector data in AI applications.

Leveraging the team's exceptional research and development capabilities, Penglai Zhishu has developed a series of AI solutions—including an AI Entertainment Recommendation System and an AI Government Text System—building on existing proprietary products. These innovations expand application scenarios and deliver comprehensive products and solutions to customers.