Alibaba Cloud officially released Hologres 4.0 in September 2025, introducing HSAP 2.0 (Hybrid Search/Analytics Processing), a unified architecture that consolidates OLAP, full-text search, vector search, and AI functions into a single engine. The release targets enterprises managing multimodal data combining text, images, logs, and sensor signals without the complexity and cost of stitching together multiple specialized systems.
What’s New in Hologres 4.0
Hologres 4.0 debuts HSAP 2.0, branded as a “hexagonal warrior” for analytics, capable of handling six workload types: OLAP, point lookups, full-text search, vector search, time-series processing, and wide-table queries. Alibaba Cloud describes it as an “all-in-one multimodal analytics and search platform” that processes data ingestion, AI transformation, and cross-modal querying via a single SQL statement.
The platform natively integrates AI Functions built-in operators like ai_embed() for vectorization and ai_gen() for image-text summarization invokable directly in SQL without custom functions or external services. These leverage Alibaba Cloud’s shared GPU pool and Qwen large language models for tasks like sentiment analysis, translation, and data masking.
Hologres 4.0 also supports automatic file synchronization from Object Storage Service (OSS), enabling direct queries on unstructured files like images and PDFs without migration into the warehouse.
Why It Matters
Traditional data architectures force developers to layer separate engines for structured analytics, search, and vector operations, creating data silos, sync delays, and high operational costs. Enterprises running AI workloads such as autonomous vehicles analyzing VIN data, CAN bus logs, and trajectory images previously needed manual result stitching across systems, slowing queries and complicating logic.
Hologres 4.0 eliminates this fragmentation by unifying scalar, text, and vector data under one engine, reducing infrastructure overhead while accelerating hybrid queries like “find vehicles with battery >40°C AND images containing crosswalks”. The serverless architecture enables dynamic scaling without pre-provisioning, further lowering costs.
Evolution from Earlier Versions
- Hologres 1.0 (2020): Introduced HSAP 1.0, merging OLAP and key-value lookups
- Hologres 2.0: Added elastic scaling and native JSONB columnar storage
- Hologres 3.0 (Sept 2024): Enabled lakehouse integration with MaxCompute, Apache Iceberg, and Paimon
- Hologres 4.0 (Sept 2025): Adds vector search, full-text retrieval, and AI Functions natively
What’s Next
Hologres 4.0 is now generally available with serverless instances in beta. Alibaba Cloud positions the platform as foundational infrastructure for AI-native applications, with continued enhancements to multimodal processing capabilities expected in future updates. Enterprises can access Hologres through Alibaba Cloud’s global regions, with integration for PostgreSQL-compatible workflows.
Featured Snippet Boxes
What is Hologres 4.0?
Hologres 4.0 is Alibaba Cloud’s unified analytics platform released in September 2025, combining OLAP, vector search, full-text search, and AI functions in one SQL-native engine via HSAP 2.0 architecture.
How does HSAP 2.0 improve multimodal analytics?
HSAP 2.0 eliminates the need for multiple specialized engines by processing structured data, text, and vectors in a single system, reducing latency and operational complexity for cross-modal queries.
What AI capabilities does Hologres 4.0 include?
It offers built-in AI Functions like ai_embed() for vectorization, ai_translate() for language conversion, and ai_gen() for content generation, all invokable via SQL without external integrations.
Can Hologres 4.0 query unstructured files directly?
Yes, Hologres 4.0 supports direct queries on images, PDFs, and other unstructured files stored in OSS through table-like interfaces, with automatic metadata synchronization and no data migration required.
