Skip to content
Percona Software for MongoDB Documentation
Rate this page
☆ ☆ ☆ ☆ ☆
Thanks for your feedback
Thank you! The feedback has been submitted.

Get free database assistance or contact our experts for personalized support.

Get help from Percona

Percona Search for MongoDB 1.70.5-3 (2026-10-29)

Installation

We are pleased to announce that Percona Search for MongoDB 1.70.5-3 is Generally Available (GA) and ready for production use. This release also adds Hugging Face Inference API as a new automatic embedding provider, giving you a hosted option to generate embeddings with open embedding models, without running your own embedding server.

This release of Percona Search for MongoDB is based on mongot 1.70.5, bringing the latest upstream improvements and fixes. For a complete list of changes, see the MongoDB release notes .

Release highlights

Percona Search for MongoDB is now Generally Available

With version 1.70.5-3, Percona Search for MongoDB graduates from technical preview to Generally Available (GA). It is now production-ready and supported for use in production environments.

You can now run full-text search with $search and $searchMeta, vector search with $vectorSearch, and automatic embedding on your production Percona Server for MongoDB deployments. To get started, see Install Percona Search for MongoDB.

Automatic embedding with the Hugging Face Inference API

Percona Search for MongoDB now supports automatic embedding with the hosted Hugging Face Inference API . The new HUGGINGFACE_INFERENCE provider joins VOYAGE and OPENAI_COMPATIBLE as a supported embedding provider.

mongot calls the Hugging Face hf-inference feature-extraction pipeline, which serves popular open sentence embedding models such as BAAI/bge-small-en-v1.5. You can use automatic embedding without running your own embedding server. For credit requirements, see Free tier and billing. mongot authenticates with your Hugging Face access token, sent as a bearer token.

By setting a custom providerEndpoint, you can also point the same provider at a dedicated Hugging Face Inference Endpoint or at a self-hosted Text Embeddings Inference (TEI) server.

As with the other providers, you configure Hugging Face models in mongot and reference them from an autoEmbed Vector Search index. mongot generates and updates embeddings for your documents as they change and creates an embedding for the query text at search time. You can configure Voyage AI, OpenAI-compatible, and Hugging Face models on the same mongot instance. The model specified in an autoEmbed index determines which provider mongot uses.

Note

The HUGGINGFACE_INFERENCE provider supports embedding tasks only. Other Hugging Face Inference API tasks, such as classification or text generation, aren’t supported.

For setup instructions, see Configure automatic embedding with the Hugging Face Inference API.

Changelog

New Features

  • PS4M-17: Percona Search for MongoDB now supports automatic embedding with Hugging Face Inference API embedding models.

Join us in shaping the future of Percona Search for MongoDB

General Availability is a milestone, not the finish line. Your real-world experience and feedback are invaluable as we continue to evolve Percona Search for MongoDB. We encourage you to run it in your environments and share your insights:

  • What challenges did it solve for you?
  • What features would you like to see next?
  • How can we make it even better for your specific use cases?

We’d love to hear from you. If you have questions or feedback, contact your Percona representative or connect with the community on the Percona Community Forum.