Velir Announces Search API Solr Dense Vector Module for Drupal 11

Velir released the Search API Solr Dense Vector module for Drupal 11 on 9 June 2025. Created by Kevin Quillen, the module leverages Solr 9.6 or higher to store and query dense vector fields within a Drupal search index. By converting content into numerical embeddings with AI providers such as OpenAI, it delivers search results based on semantic similarity rather than simple keyword matches.

The module builds on the Drupal Search API and AI frameworks. Site administrators can choose an embedding provider and model, set vector dimensions, and select a similarity function through a dedicated configuration interface. During indexing, content fields become dense vectors sent to Solr. At query time, user inputs are transformed into vectors that Solr compares to rank results by meaning.

Required components include Solr 9.6 or higher, Search API Solr 4.x, the Drupal AI module, and a configured embedding service. Current Solr limitations permit only one dense vector field per index.

Early use cases include retrieval-augmented generation workflows, smarter content discovery, and potential chatbot support. Velir plans to roll out the module on client projects throughout 2025 and gather feedback for future enhancements, such as result-set summaries and vector-management tools.

Full details, installation steps, and source code are available on the Drupal.org page.