BM25 vs Gemini vs Vector vs Hybrid Search

BM25 is a ranking algorithm built into SQLite via its FTS5 engine. It scores documents by term frequency weighted by inverse document frequency. Runs locally in milliseconds.

Gemini reads the full document and reasons about your query to produce natural-language answers. Powerful but requires a network round-trip.

Vector Search uses Gemini embeddings + LanceDB cosine similarity. Captures semantic meaning beyond exact keyword matches.

Hybrid combines BM25 + Vector retrieval, deduplicates with Reciprocal Rank Fusion, reranks with Gemini, and generates a cited answer. The best of all worlds.