Meilisearch
A lightning-fast search engine API bringing AI-powered hybrid search to your sites and applications.
Last verified:
What is Meilisearch?
Meilisearch is a lightning-fast, open-source search engine designed for developers who want to add powerful search functionality to their websites and applications. It serves as a unified search and AI retrieval platform that delivers search-as-you-type results in under 50 milliseconds. The engine is trusted by over 20,000 teams worldwide including major companies like Hugging Face and Louis Vuitton.
Key features include AI-powered hybrid search that combines full-text keyword search with semantic vector search, typo tolerance that handles misspellings automatically, semantic search for deeper query understanding, multimodal search across images/video/audio alongside text, filtering and faceting for complex search interfaces, geosearch for location-based results, vector storage for RAG applications, federated search across multiple data sources, and comprehensive search analytics. The platform offers plug-n-play deployment with smart presets requiring zero configuration, supports JSON/CSV/NDJSON data formats, and provides SDKs for 10+ programming languages.
Meilisearch is ideal for developers building e-commerce search, documentation search, app search, website search, AI retrieval systems, and RAG applications. It's particularly suitable for startups and teams who want powerful search without the complexity of Elasticsearch or Solr. The platform serves both self-hosted use cases (MIT licensed Community Edition) and managed cloud deployments with automatic scaling, backups, and updates.
Meilisearch pricing
Pricing model: Freemium
Meilisearch offers two main options: Self-hosted (Open Source) is completely free under MIT license with unlimited indexes, unlimited documents, and all search features including AI hybrid search. Meilisearch Cloud starts at $20/month with a 14-day free trial requiring no credit card. The Cloud Build plan is $30/month (100K docs, 50K searches, 2GB RAM, 10GB storage, managed hosting, auto backups). The Cloud Pro plan is $300/month (8GB RAM, 50GB storage, priority support, SLA guarantee). Resource-based Starter plan is $23/month (0.5 vCPU, 1GB RAM, 32 GiB storage included). Additional costs: $0.165 per extra GiB storage, $0.15 per GB outbound bandwidth. Email support included with all Cloud plans. Enterprise custom pricing available.
Meilisearch pros
- Lightning-fast sub-50ms search-as-you-type responses
- Out-of-the-box excellent relevancy without configuration
- Smart presets enable zero-configuration setup
- Deploy in minutes rather than hours
- Typo tolerance handles misspellings automatically
- Open-source MIT license for self-hosting
- AI-powered hybrid search combining keyword and semantic search
- SDKs available for 10+ programming languages
- Self-explanatory API designed for developer experience
- Unlimited indexes and documents when self-hosted
- Supports multimodal search across text, images, video, and audio
- Built-in filtering, faceting, and sorting capabilities
- Geosearch for location-based filtering and sorting
- Vector storage for advanced similarity queries and RAG
- Federated search across multiple data sources simultaneously
- Comprehensive search analytics for data-driven decisions
- Scales seamlessly from prototype to production
- Trusted by 20,000+ teams including Hugging Face and Louis Vuitton
Meilisearch cons
- Cloud plans start at $23-30/month with no permanent free tier for managed service
- Self-hosting requires managing your own servers, updates, and backups
- Maximum 10 query terms for full-text search (configurable but limited)
- Default maximum 1,000 results per search (configurable)
- Recommended index size under 2 TiB despite 80 TiB theoretical max
- Hardware requirements need 10x disk space of raw dataset
- Anonymous telemetry collection by default (requires opt-out)
- Email support only included (no phone or chat support)
- No annual discount information publicly listed
Frequently asked questions about Meilisearch
I have never used a search engine before. Can I use Meilisearch anyway?
Yes! No knowledge of Elasticsearch or Solr is required. Meilisearch is designed to be easy to use and accessible to all developers. The fastest way to start is with Meilisearch Cloud, which gives you a running instance in minutes. You can also self-host Meilisearch if you prefer. SDKs are available for many languages and frameworks to help integrate Meilisearch into your project.
Should I use Meilisearch Cloud or self-host?
Meilisearch Cloud is recommended for most users. It handles provisioning, scaling, backups, and updates automatically, and includes built-in analytics and monitoring. Self-hosting gives you full control over infrastructure and is available under the MIT license (Community Edition), but requires managing your own servers, updates, and backups. Cloud removes infrastructure management entirely with fully managed service and automatic upgrades.
How does Meilisearch compare to other search engines?
Meilisearch maintains detailed comparisons with Elasticsearch, Algolia, Typesense, and others. Meilisearch is simpler to use than Elasticsearch or Solr, requires no complex configuration, and delivers better out-of-the-box relevancy. Unlike Algolia, Meilisearch is open-source with truly free self-hosting. Compared to Typesense, Meilisearch offers AI-powered hybrid search and more predictable pricing starting at $30/month flat.
What are Meilisearch's limits?
Key limits include: approximately 4.3 billion maximum documents per index, approximately 80 TiB maximum index size (recommended under 2 TiB), 65,536 maximum attributes per document, 10 maximum query terms for full-text search (configurable), and 1,000 default maximum results per search (configurable). For the full list, see known limitations in the documentation.
How do I update Meilisearch?
Meilisearch Cloud instances can be updated with one click from the Cloud dashboard. For self-hosted instances, the update guide covers dumpless upgrades for version 1.13+ and dump-based migration for older versions. Zero-downtime upgrades are available.
Can I use Meilisearch for multi-tenant applications?
Yes. Meilisearch supports multitenancy with tenant tokens, which let you control which documents each user can search without maintaining separate indexes. This enables secure per-user or per-tenant search access within a single index.
What are the hardware requirements for self-hosting?
Requirements depend on dataset size, number of searchable/filterable fields, and query volume. As a starting point, provision a machine with at least ten times the disk space of your raw dataset. More RAM means faster searches since search speed depends on the RAM-to-database-size ratio. More searchable/filterable fields and ranking rules increase database size. More CPU cores allow handling more concurrent search queries.
Is killing a Meilisearch process safe?
Yes. Killing Meilisearch is safe even during indexing. When you restart, it resumes the task from the beginning. This applies to all operations including document indexing. See the asynchronous operations guide for more details on tasks and batches.
I keep getting a 400 Bad Request when adding documents. What's wrong?
This usually means your data is not in a valid format. Common causes include extraneous commas, mismatched brackets, or missing quotes. Meilisearch accepts JSON, CSV, and NDJSON formats. When adding or replacing documents, you must enclose them in an array even if there is only one document.
Is there a public roadmap?
Yes. Visit the public roadmap at roadmap.meilisearch.com to see planned features and ongoing work. You can also join the community on GitHub to open issues and PRs, request new features, and vote on the ones that matter most to you. The Discord community is available for asking questions.