Trusted by leading companies
How RAG works
Retrieval-Augmented Generation combines search with AI to deliver accurate, grounded answers.
User asks a question
Natural language input from your chat interface.
Meilisearch retrieves context
Hybrid search finds the most relevant documents instantly.
LLM generates answer
AI synthesizes a response grounded in your actual content.
"Based on your documentation, the recommended approach is to use the /search endpoint with the following parameters…"
Everything RAG needs
A complete retrieval foundation for AI applications. No assembly required.
Vector storage
Store embeddings alongside your documents. No separate vector database needed.
- Embeddings handled for you
- Scales to millions of docs
- 8+ AI providers
Hybrid retrieval
Combine keyword precision with semantic understanding for better context.
- Tunable keyword/semantic mix
- Best of both approaches
- Results under 50ms
Context formatting
Control exactly what content gets sent to your LLM.
- Pick which fields to send
- Use simple templates
- Include any metadata
The complete RAG toolkit
Everything you need to build AI applications with accurate, grounded responses.
Vector search
Store and query embeddings alongside your data. Stays fast at millions of documents.
Hybrid retrieval
Combine keyword and semantic search to fetch the right context. Better inputs lead to better AI answers.
Similar documents
Surface the closest matches to any item with one call. Perfect for "more like this" and related content.
Control what the LLM sees
Choose exactly which fields go into each prompt. Cleaner context produces sharper answers.
LangChain retriever
First-class integration. Drop into existing RAG pipelines with one import.
MCP server
Model Context Protocol for AI agents. Tool-using assistants with live data.
Fits your AI stack
First-class integrations with the tools you already use.
LangChain
Official retriever integration for Python and JavaScript.
from langchain_meilisearch import MeilisearchRetrieverLlamaIndex
Use as a vector store in your LlamaIndex pipelines.
from llama_index.vector_stores import MeilisearchVectorStoreMCP server
Model Context Protocol for Claude and other AI agents.
npx @modelcontextprotocol/server-meilisearchWorks with all major LLM providers
Native integrations for OpenAI, Anthropic, Mistral, Google, and more.
58 models from 13 providers
OpenAI
6 models
Anthropic
3 models
5 models
Mistral AI
7 models
Cohere
2 models
DeepSeek
3 models
AWS Bedrock
4 models
Hugging Face
6 models
Ollama
8 models
Together AI
4 models
Fireworks AI
4 models
Cloudflare AI
4 models
Moonshot AI
2 models
Custom
Any provider
Meilisearch is compatible with any model offering a REST API and tool calling capabilities.
Built for AI applications
The retrieval layer for any AI-powered experience.
Chatbots & assistants
Grounded answers from your actual content.
Semantic search
Find documents by meaning, not just keywords. "budget phones" finds "affordable devices".
Recommendations
Similar document API for "more like this" features. Products, articles, content.
Question answering
Extract answers from documents with source attribution. Full transparency.
Content discovery
Help users explore related content based on what they're viewing.
AI agents
Give tool-using AI assistants access to your data via MCP.