Features
Auto-scaling compute and storage
Tiered storage (DRAM, SSD, object) for cost efficiency
Vector similarity search (cosine, Euclidean, IP, etc.)
Hybrid search (dense, sparse, multimodal)
Tunable consistency levels
One-click data migration to dedicated or open-source Milvus
Backup and restore (scheduled, cross-region)
Integration with embedding models (OpenAI, Cohere, Hugging Face)
RESTful API and SDKs (Python, Java, Go, Node.js)
Metric monitoring and alerts
Access controls (RBAC, project/user management)
Encryption at rest and in transit
On-demand compute for lake-scale search jobs (zero-copy external data)
Support for collections, indexes, and filtering
99.95% uptime SLA (Enterprise tier)
Fully managed vector database
Automatic index tuning and upgrades
Writes acknowledged in under 100ms
Consistent p99 latency at any scale
Metadata filtering during queries
Namespace-per-agent isolation
Dense, sparse, and full-text indexes
Hybrid search (vector + BM25) via Pinecone Nexus public preview
Integrated embedding and reranking (Cohere Rerank 4.0 Fast GA)
Customer Managed Encryption Keys (CMEK)
Open-source monitoring stack (June 2026)
Multi-region and multi-cloud support (AWS, GCP, Azure) on Builder+
