Free Lesson
Introducing Luxir: Can we build the fastest search engine?
60 min
Oct 26, 2026 12:00 PM
What you'll learn
Why Solr’s creator is building a new search engine
What decades of building Lucene and Solr taught Yonik Seeley, and why he chose to start fresh with a clean-slate design.
The Vision: the fastest, native engine for modern retrieval
How Luxir enables hybrid BM25, filtered vector search, & exact analytics, optimized with native code for the AI era.
Optimizing for the most search per core, GB, and dollar
Work-stealing across every core, memory-mapped segments, SIMD scoring, and why it scales up before scaling out.
Early benchmarks, and what’s next for Luxir
See the first performance numbers, what’s shipped vs. still coming, and how to try Luxir or get involved.
Why this topic matters
The original creator of Solr just launched a new open source search engine named Luxir. It's a clean-slate design and implementation... but WHY? Today's most widely used search engines run on the JVM and scale by adding more nodes, but you pay for that inefficiency in the cloud every month. As agents and AI apps multiply query volumes, cost per query matters more than ever. Come hear the project vision and see early benchmarks from Yonik himself!
