You're viewing version 2.19 of the OpenSearch documentation. For the latest version, see the current documentation. For information about OpenSearch version maintenance, see Release Schedule and Maintenance Policy.

Vector search API

In OpenSearch, vector search functionality is provided by the k-NN plugin and Neural Search plugin. The k-NN plugin provides basic k-NN functionality, while the Neural Search plugin provides automatic embedding generation at indexing and search time.

For k-NN plugin APIs, see k-NN API.

In addition to plugin-specific APIs, the following APIs support vector search functionality:

• k-NN vector
• k-NN query
• Neural query
• Neural sparse query
• Ingest pipelines
• Ingest processors:
  • ML inference
  • Sparse encoding
  • Text chunking
  • Text embedding
  • Text/image embedding
• Search pipelines
• Search processors:
  • ML inference (request)
  • ML inference (response)
  • Neural query enricher
  • Neural sparse query two-phase
  • Normalization
  • Rerank
  • Retrieval-augmented generation
  • Score ranker