What’s New in Milvus 2.3
We are thrilled to announce the release of Milvus 2.3.0. This release contains many exciting new features and improvements. This blog post will highlight some of the more prominent features. For a complete list of changes, be sure to check the release notes.
📦 PyPI: https://pypi.org/project/milvus/
📚 Docs: https://milvus.io/docs
🛠️ Release Notes: https://milvus.io/docs/release_notes.md#v230
🐳 Docker Image: docker pull
🚀 Release: milvus-2.3.0
🖥️ Computational upgrades with GPU & ARM64:
GPU Indexing via Nvidia: QPS has improved over threefold compared to the previous CPU HNSW indexing. For computation-intensive datasets, there's almost a tenfold increase in performance.
ARM64 Docker Image: ARM CPUs deliver a 20% performance boost while being 20% more cost-effective.
🔎 Search & indexing enhancements:
Range Search: A more granular search mechanism enabling vector retrieval based on defined distances from an input vector, making proximity-based data retrievals more intuitive.
ScaNN Index Integration: Compared to HNSW or IVFFlat, the ScaNN index provides a 20% performance improvement and is seven times faster than IVFFlat.
Growing Index: As data streams in, Milvus 2.3 indexes it in real-time, ensuring quicker query executions.
🔀 Data pipeline tools:
Iterator in Pymilvus: Enables batch-export of large quantities of vectors, especially beneficial when handling data in the tens of thousands.
Upsert Operation: Simultaneous update or insert of data in a single request, streamlining data management and enhancing efficiency.
CDC Support: Facilitates data syncing across data centers, allows incremental backups, and ensures smooth data migration.
We've also improved the system's operability, load balancing, and query performance. Several bugs have been addressed, and we've optimized core components such as the ‘QueryNode’. Our message queue system has been overhauled for better efficiency. Additionally, tools like Birdwatcher and Attu have seen updates for better functionality. Dive into the Milvus 2.3 release notes for a comprehensive overview.
Closing notes:
A big thank you to our developer community. Your contributions, PRs, and feedback have shaped Milvus 2.3. We're keen to see how you integrate these updates and await your continued feedback. Keep coding! 🖥️🔧
Steffi LiSteffi is the Director of Product Marketing at Zilliz, with prior experience leading the GTM strategies for open-source data technologies such as Apache Kafka and Apache Airflow. Passionate about travel, yoga, and non-fiction reading, she brings a balanced and enriched worldview to her professional pursuits.
Keep Reading
Notion's Vector Search Is Excellent. Their Next Problem Is Harder.
Notion solved vector search scaling in two years. The next bottleneck — offline context engineering, unified data, and the real-time/offline gap — is harder.
How Zilliz Ended Up at the Center of NVIDIA’s Unstructured Data Story at GTC 2026
If unstructured data is the context of AI, then the ceiling of AI applications will be set not just by models, but by how mature the infrastructure for unstructured data becomes.
8 Latest RAG Advancements Every Developer Should Know
Explore eight advanced RAG variants that can solve real problems you might be facing: slow retrieval, poor context understanding, multimodal data handling, and resource optimization.