Manticore Search

Asked for merino wool socks. Got pants. Why? We tested six ways to search the same 82,524 products in Manticore Search Playground. Explore what works and where search goes wrong. mnt.cr/go/fFij6E

What’s the p95 latency? What share of requests finish within 200 ms? How widely do values vary? You can now answer these questions directly in Manticore — we show you how with SQL and JSON queries. mnt.cr/go/xgRjUH

We replaced buffered vector reads with mmap in Manticore’s columnar storage. On DBpedia, KNN throughput rose 2.57–3.13×, reaching 85–92% of row-wise performance. Here’s why stable pointers matter: mnt.cr/go/S3sNTv

Less load on MySQL, no dedicated search servers. Kiva uses Manticore Search to power search across customer databases with near-zero extra infrastructure cost. Here’s how: mnt.cr/go/yottqQ

Better search without replacing MariaDB: wine.co.za uses Manticore Search across ~145,000 records, with daily indexing, flexible queries and suggestions as visitors type. Here’s how its .NET setup works: mnt.cr/go/yiXmgu

Bookbot uses Manticore Search to match newly available books against 3.5 million saved searches. Its Watchdog alerts account for roughly 10% of revenue, according to Bookbot. Here’s how it works: mnt.cr/go/Yq1rML

Vector search missing answers in long documents? Manticore Search can now split text into chunks and embed each one automatically. Search every chunk, get one result per document. mnt.cr/go/MV6uBV

One vector per document is not always enough. Manticore Search 29.9.0 adds chunked auto-embeddings: split long documents into meaningful chunks, store multiple vectors per document, and find relevant passages even deep inside the text. mnt.cr/go/zVQu9s

Manticore Search 29.9.0 is out 🚀 Long documents no longer have to fit into one embedding: split them into chunks, keep multiple vectors per document, and search each chunk independently with KNN. Also: mmap columnar access by default + lots of fixes. mnt.cr/go/eCDW2F

OpenSearch Dashboards + Manticore Search 🤝 Explore Manticore data with Discover, build visualizations, and create interactive dashboards — while Manticore handles search and analytics on the backend. See how to set it up: manticoresearch.com/blog/man…

Need the latest N items from each group without returning N rows per group? In Manticore Search 28.6.6, GROUP_CONCAT() can now ORDER BY + LIMIT inside the aggregate: one row per group, total count intact, top IDs sorted. manticoresearch.com/blog/gro…

Searching long hashes, message IDs, or email addresses in Manticore Search? dict='keywords_32k' raises the token limit from 42 bytes to 32 KB, making full-text, prefix, and infix search practical for long machine identifiers. New guide 👇 manticoresearch.com/blog/dic…

How does Serbia's largest classifieds marketplace search 5.6M active ads in ~10 ms? KupujemProdajem runs Manticore Search at ~150 queries/sec, with real-time updates, replication, custom ranking — and hybrid search next. Read the case study 👇 manticoresearch.com/blog/kup…

Full-text search isn't dead. It's just not enough anymore. Users don't want 10 blue links or 20 product cards. They want to ask: "Find me black waterproof running shoes under $120 and tell me which are best for daily use." We built Conversational Search in Manticore to do exactly that. And benchmarked it. 👇 manticoresearch.com/blog/con…

🔐 New interactive Manticore Search course: Authentication! In just 10 minutes learn how to enable auth, bootstrap an admin, manage users & fine-grained permissions, and use HTTP Basic Auth + bearer tokens. 👉️️️️️️ play.manticoresearch.com/aut…

🚀 The latest Manticore Search features are now available across all our official clients: 🔐HTTP Basic & Bearer authentication 💬 Conversational search 🔎 Hybrid search 🆔 UUID document IDs 📊 New aggregations & facet_filter_mode manticoresearch.com/blog/man…

UUIDs as document IDs in Manticore Search 🧩 A hands-on guide to table setup, CRUD via SQL and JSON API, batch loading with /bulk, UUID generation and validation, columnar RT tables, and replication. 👇 manticoresearch.com/blog/uui…

🚀Cool update in Manticore Search! Now you can use UUID as document IDs — one single ID for your primary database and Manticore. No more mapping tables! 👉 manticoresearch.com/blog/uui…

🚀 Manticore Search 28.6.6 is out. • UUID document IDs for RT tables • Ordered, limited GROUP_CONCAT() • 16 fixes across backups, replication, query handling, SQL compatibility, and secondary indexes 👉 manticoresearch.com/blog/man…

🚀 New in Manticore Search: Faceted search without the manual filter pain! Tired of rebuilding nearly identical queries for every facet once users apply filters? facet_filter_mode (strict / auto / max) now handles selected, available & unavailable options right in the API. Perfect for e-commerce filter panels 👉 manticoresearch.com/blog/fac…