Amazon OpenSearch Service adds GPU-accelerated and auto-optimized vector indexes

This article announces GPU-acceleration and auto-optimization features for Amazon OpenSearch Service vector indexes.

• Build billion-scale vector databases in under an hour with GPU-acceleration
• Auto-optimize vector indexes for search quality, speed, and cost trade-offs
• Serverless jobs automatically evaluate k-NN algorithms, quantization, and engine settings
• GPU-accelerated indexing up to 10X faster at quarter of current cost
• Serverless GPUs activate dynamically; billing only when used
• Supports semantic search, recommendation engines, and agentic systems
• GPU-acceleration available in five regions; auto-optimize in nine regions

These features significantly reduce time and expertise needed to build optimized large-scale vector databases, enabling faster AI application innovation.