What is Meilisearch?
Meilisearch is a search API that developers use to add fast, relevant search to websites and applications. It handles full-text search, semantic search, hybrid search, and vector storage in one platform. Teams use it to power product search, knowledge bases, documentation portals, and AI applications. The platform runs as a cloud-hosted service or as a self-hosted open-source engine. SDKs are available for JavaScript, Python, PHP, Ruby, Java, Go, .NET, Rust, Swift, and Dart.
Meilisearch Video
Features & Benefits
- Full-Text Search: deliver keyword-based search with sub-50ms response times. Includes typo tolerance, search-as-you-type, prefix matching, and relevance ranking.
- Typo Tolerance: find the right result even when users misspell. Configure strictness per field so exact-match fields stay precise.
- Relevance Tuning: reorder ranking rules or apply custom scoring logic such as popularity or stock level.
- Word Intelligence: split and join compound words automatically so “newspaper” and “news paper” resolve to the same results.
- Deduplication: surface one best match per item even when many variants exist in the index.
- Semantic Search: match queries by meaning and intent rather than exact keywords using AI-generated embeddings. Understands that “affordable housing” relates to “budget apartments.”
- Hybrid Search: blend keyword precision with semantic understanding in a single query. Tune the keyword-to-semantic ratio per use case without code changes.
- Multimodal Search: search across images, video, and audio alongside text using AI-powered embeddings.
- Federated Search: query multiple indexes in one API request and receive a single server-ranked result list. Supports per-source boosting and unified pagination.
- Faceted Search and Filtering: build complex search interfaces with filters, facets, and sorting. Compatible with full-text and semantic search.
- Geosearch: filter and sort results by location.
- Vector Storage: store and retrieve embeddings alongside documents for similarity queries and RAG applications. Scales to millions of documents.
- RAG Infrastructure: retrieve grounded context for LLM applications using hybrid retrieval. Control which fields feed the prompt. Integrates with LangChain and LlamaIndex.
- MCP Server: give tool-using AI agents live access to indexed data via Model Context Protocol.
- Similar Documents: surface the closest matches to any document in one API call.
- Search Analytics: gain data-driven insights from search behavior.
- Embedding Provider Integrations: connect to OpenAI, Mistral, Cohere, Google, HuggingFace, Ollama, Voyage, Jina, Together AI, Cloudflare AI, and others. Bring any model with a REST API.
- Multilingual Support: handle 30-plus languages with automatic detection and language-aware tokenization.
What can Meilisearch do?
- Add search to a website
- Add search to a mobile app
- Build a product search for an ecommerce site
- Search across multiple indexes in one query
- Search documents by meaning not keywords
- Combine keyword and semantic search in one request
- Store vector embeddings alongside documents
- Build a RAG application with hybrid retrieval
- Give an AI agent access to a search index via MCP
- Find similar documents by vector similarity
- Filter search results by location
- Search images and video with text queries
- Build a knowledge base search
- Handle typos in search queries
Real-World Applications
An ecommerce team building product search may find that a search API handling both keyword and semantic queries reduces zero-result searches. When a shopper types “comfortable office footwear” and the catalog uses “ergonomic work shoes,” semantic matching in the search API bridges that gap. Filters, facets, and geosearch can layer on top to narrow results by price, brand, or store proximity.
SaaS products that need to surface content across products, help articles, and FAQs can use the federated search API to query all three in a single request. The server merges and ranks results before returning them. The team avoids writing client-side merging logic and gets cross-index relevance out of the box.
Teams building AI applications may use the platform’s RAG infrastructure to give an LLM accurate, current context from their own data. Hybrid retrieval pulls the most relevant documents. Field-level controls let engineers decide exactly what content enters each prompt. LangChain and LlamaIndex integrations drop into existing pipelines with a single import.
Developers at open-source projects or startups who want to self-host can run the engine without a cloud dependency. The same search API runs locally or in production. SDKs across ten-plus languages mean the integration fits most existing stacks without significant retooling.
