Manticore Search 13.11.0: Introducing Auto Embeddings and Enhanced AI Search – ██FR█████ █INTELL███████████

This content originally appeared on DEV Community and was authored by Sergey Nikolaev

We’re excited to announce the August 2025 release of Manticore Search 13.11.0, a major update featuring Auto Embeddings — our new way of making AI-powered semantic search simple and efficient. This version also includes automatic embedding generation, bug fixes, and several improvements.

Auto Embeddings: AI Search Made Easy

The highlight of Manticore Search 13.11.0 is Auto Embeddings — a game-changing feature that makes semantic search as easy as SQL. No need for external services or complex pipelines: just insert text and search with natural language.

What Auto Embeddings Offer:

Automatic embedding generation from your text

Natural language queries that understand meaning, not just keywords

Support for multiple models (OpenAI, Hugging Face, Voyage, Jina)

Smooth integration with SQL and JSON APIs

Quick Example

-- Create table with auto-embeddings
CREATE TABLE products (
    title TEXT,
    description TEXT,
    vector FLOAT_VECTOR KNN_TYPE='hnsw' HNSW_SIMILARITY='l2'
        MODEL_NAME='sentence-transformers/all-MiniLM-L6-v2'
        FROM='title,description'
);

-- Insert data (embeddings generated automatically)
INSERT INTO products(id, title, description) VALUES
  (1, 'wireless headphones', 'Bluetooth headphones with noise cancellation'),
  (2, 'hiking backpack', 'Lightweight backpack for outdoor adventures');

-- Search with natural language
SELECT id, title
FROM products 
WHERE knn(vector, 3, 'portable audio device for music');

Results:

+------+---------------------+
| id   | title               |
+------+---------------------+
|    1 | wireless headphones |
...
+------+---------------------+

Here, semantic search correctly matched “wireless headphones” with the phrase “portable audio device for music,” even though no keywords overlapped.

Learn More

Want a full deep dive? Check out our dedicated article: Introducing Auto Embeddings: AI-Powered Search Made Simple

Other Improvements

Configuration

Boolean Simplify Support: Added boolean_simplify option for faster query processing
System Optimization: Sysctl config now auto-increases vm.max_map_count for large datasets
Package Management: RPM packages no longer own /run directory for better compatibility

Bug Fixes

Fixed scroll option with large 64-bit IDs
Fixed KNN crashes when using filter trees
Fixed /sql endpoint behavior (removed unsupported SHOW VERSION)
Fixed duplicate ID handling in columnar mode
Fixed crashes with joined queries using multiple facets
Fixed delete/update commits in bulk transactions
Fixed crashes when joining on non-columnar string attributes

System & Integration

Updated Windows installer script
Fixed local time zone detection on Linux
Improved JDBC+MySQL driver compatibility with transaction_read_only
Enhanced error reporting across components
Improved master-agent communication for embeddings

Compatibility

Manticore Search 13.11.0 is fully backward compatible:

No breaking changes in standard use cases
Smooth upgrades from 13.x versions
Auto Embeddings work alongside current search features
APIs are extended, not replaced

Everything is designed to work seamlessly with your existing data and queries.

Upgrade

To upgrade, follow the installation guide.

Ready to try Auto Embeddings? Start with the documentation.

Need help or want to connect?

Join our Slack
Visit the Forum
Report issues or suggest features on GitHub
Email us at contact@manticoresearch.com

For full details, see the Changelog.

This content originally appeared on DEV Community and was authored by Sergey Nikolaev