New Performance Record! GBase 8c Tops the HyBench Benchmark

In the latest HyBench benchmark test, GBase 8c, the next-generation AI-native database independently developed by GBASE, achieved a H-SCORE of 3810.36 — setting a new record and securing the No. 1 ranking. As an authoritative benchmark for HTAP scenarios, HyBench certification validates GBase 8c's industry-leading technical capabilities in hybrid transaction and analytical processing.

HyBench is China's first HTAP database benchmark based on online transaction risk control scenarios. Since its launch, it has served as an industry benchmark. Developed jointly by the China Software Testing Center, Tsinghua University, and multiple database vendors, HyBench is designed to reflect real-world application scenarios for HTAP databases. Its data model is built around online financial transaction analytics, supporting three typical HTAP workloads: OLTP, OLAP, and OLXP. Today, HyBench has become an authoritative measure for validating the HTAP capabilities of commercial products, open-source solutions, and integrated offerings in the database field.

In the latest HyBench benchmark test, GBASE's GBase 8c database management system completed the test with outstanding results, setting a new record and claiming the top spot. Notably, GBase 8c demonstrated HTAP performance far surpassing that of other distributed transaction-processing databases — achieving nearly 5× the score of the previous record holder. This leap in performance stems from GBase 8c's proprietary AI-Native Hybrid Execution Engine, which uses built-in reinforcement learning models to enable microsecond-level workload-aware execution path switching within the query optimizer. Combined with a lightweight GTM (Global Transaction Manager) that replaces traditional 2PC blocking commits with logical clock asynchronous validation, the engine fundamentally eliminates resource contention between OLTP and OLAP workloads and minimizes cross-node coordination latency — delivering near-linear throughput scaling under mixed workloads.

GBase 8c is GBASE's next-generation AI-native database, with its core capabilities best summarized as "multi-model, multi-form" — supporting multiple storage models and flexible deployment forms.

Multi-Model Storage: One System, Multiple Data Engines

GBase 8c supports multiple storage modes, including row-store, column-store, GIS, time-series, and vector. Row-store is optimized for OLTP scenarios, supporting high-concurrency transactions through MVCC and WAL logging. Column-store is designed for OLAP analytics, delivering high compression ratios and low I/O overhead. Row-column hybrid storage allows a single table to have both physical forms simultaneously, with the query optimizer automatically selecting the optimal execution path for OLTP and OLAP workloads.

In addition to supporting traditional OLTP, OLAP, and HTAP workloads, GBase 8c natively supports vector data storage, enabling hybrid storage and retrieval of vector and scalar data. Unified multi-model storage makes "one copy of data, multiple access methods" a reality, effectively eliminating data silos.

Multi-Form Deployment: Flexible Adaptation, Elastic Scaling

GBase 8c offers three deployment forms: primary-standby, distributed, and storage-compute separation.

Primary-standby deployment — one primary with multiple standbys, suitable for scenarios with lower data volumes where extreme single-node performance is required.

Distributed deployment — fully redundant components with high availability and storage-compute separation, allowing horizontal scaling of compute and storage independently.

Storage-compute separation deployment — built on shared object storage with a stateless compute layer, supporting second-level scaling. During idle periods, compute costs approach zero, providing robust support for massive concurrent access from AI agents.

Broad Ecosystem Compatibility
Built on its multi-model, multi-form architecture, GBase 8c delivers extensive ecosystem compatibility — fully compatible with PostgreSQL syntax and highly compatible with common syntax from Oracle, MySQL, and SQL Server. It adapts to mainstream domestic chips, operating systems, and middleware. The accompanying automated migration tools support full, incremental, and reverse synchronization, enabling a single database to seamlessly replace multiple heterogeneous database systems — dramatically reducing application adaptation costs.

It is precisely this "multi-model, multi-form" core architecture and broad ecosystem compatibility that enabled GBase 8c to demonstrate outstanding HTAP performance in the HyBench benchmark — delivering high-concurrency transaction processing and complex analytical queries within a single system, with excellent data freshness, ultimately securing the top ranking.

Proven in Production
Today, GBase 8c has achieved large-scale deployment across finance, telecommunications, healthcare, government, and other critical industries. From passing the national security and reliability assessment to topping the HyBench benchmark, GBase 8c continues to demonstrate solid technical capabilities, delivering secure, stable, and reliable data storage and management services for mission-critical core systems.

HyBench's No. 1 ranking confirms the technical maturity of GBase 8c in HTAP scenarios. As digital transformation accelerates across finance, energy, government, internet, and other sectors, demand for databases that deliver "one system for multiple workloads — transaction, analytics, and AI all in one" continues to grow. GBASE will take this benchmark achievement as a new starting point, continuing to deepen core database technologies in multi-model, cloud-native, and AI integration — providing autonomous, independently controlled data infrastructure to support digital and intelligent transformation across industries.