Bedrock AgentCore's Web Answer plans its own searches and returns a synthesized answer with citations. us-east-1 only, $10 per 1,000 calls, no model choice.
A Berkeley-led vector index keeps a pruned graph, not the vectors, so a local RAG index can be a few percent of the data's size. Costs and how to try it.
TokenGate, from Stanford and Berkeley researchers, trains a gate that drops unneeded context tokens early, cutting prefill time and memory on small models.
Before 1.20, any host that reached Qdrant's internal port 6335 could join a cluster and copy shards. The fix needs an API key. BM25 and turbo8 also land.
Five open NVIDIA retrieval models search page images and text, rerank results and parse PDFs, under the permissive OpenMDW license. What to check first.
Exa 3.0 adds a larger embedding model, GA highlights that use 95% fewer tokens, billions of new documents and a $4 Lite search tier. Benchmarks self-run.
DeepSeek's API guide documents Search and Fetch at 5 requests a second. The Chinese pages price each at 5 yuan per 1,000 successes; the English pages say $0.75.
Cross-encoders score a query and a document in one pass, which makes them more accurate than embedding search and too slow for a whole corpus. Use them to rerank the top hits.
Google published EmbeddingGemma 2, an open model that maps text, code, images, video, and audio into one 768-wide vector. Developers can load fewer encoders or store a shorter vector.
Chunk size, overlap and boundaries decide what a retriever can return. This post covers defaults documented by Azure AI Search and OpenAI, plus a way to test a setting on your own questions.
SQLite puts application data, full-text search and, with an extension, vector search in a local database file. Here is where that fits and when PostgreSQL fits better.
Retrieval-augmented generation is an architecture, and the vector database is one box in it. The whole loop, the RAG variants worth knowing, reranking, and when a knowledge graph beats similarity.
turbopuffer made native embedding generally available (Sep 29, 2026): embed on write, on query, or keep BYOE. Systems overlap with index metadata fetch; Linear/Readwise beta color is vendor-stated.