MindsDB Product Updates - October 2025

This month’s product update is packed with powerful upgrades that make MindsDB even more robust, flexible, and user-friendly. From major enhancements to Knowledge Bases and SQL operations to smoother GUI interactions, improved testing and integrations, and refreshed documentation, every change is designed to help you build, deploy, and manage AI workflows more efficiently.

Knowledge Bases

This update brings a series of improvements that make MindsDB’s Knowledge Bases more reliable, powerful, and easier to work with. We’ve removed several hard-coded variables and improved the internal class extraction process, making Knowledge Bases cleaner, faster, and easier to maintain for developers and enterprise users alike.

Support for Ollama and Amazon Bedrock has been added, expanding the range of LLMs you can integrate directly into Knowledge Bases. Whether you’re using local models with Ollama or leveraging Bedrock’s managed infrastructure, these new integrations make MindsDB even more adaptable to your AI workflows.

We’ve also made several enhancements to query handling and data operations. The GREATER THAN filter bug has been fixed, OpenAI handlers have been deactivated for simpler configuration, and you can now filter by chunk_content for more precise searches. Additionally, the ALTER KNOWLEDGE BASE command has been revised to better handle metadata as columns, resulting in smoother and more accurate updates.

Under the hood, we’ve fixed key unit and integration tests across Knowledge Bases, significantly improving overall performance and reliability. Our continuous integration pipeline now also includes Python versions 3.10 through 3.13, ensuring broader compatibility and stability across different environments.

Integrations

We’ve added Transformers to the main dependencies, improving MindsDB’s ability to analyze and process data inputs more efficiently. This change enhances the app’s overall performance and usability when working with advanced AI models.

A number of fixes have been made to improve integrations and reliability. The Shopify handler now has better error handling and clearer message displays, while a streaming request issue that caused a 400 Error in the Respond Tab has been resolved.

Several integrations have also been improved to deliver a better user experience. Custom URLs now work correctly in the GitLab integration, issues in Zendesk have been fixed, and Ollama support has been added to Knowledge Bases. We also addressed smaller bugs, including fixes related to Azure OpenAI, MindsDB GUI auto-restart, and data handler caching.

Connection handling has been refined for VLLM, Snowflake, and database URIs, ensuring more stable connectivity. For Windows users, a new dependency has been introduced for DuckDB, reducing compatibility issues and improving performance.

Finally, the MySQL API connection queue has been increased to handle more complex operations seamlessly, and we’re happy to share that this release — v25.10.1 — now includes full support for the Oracle handler.

Bug Fixes and Improvements

We’ve made significant improvements to our testing infrastructure. Several main unit tests have been corrected and now run automatically on every push to the main branch. Test coverage has also been expanded to include Python versions 3.10 through 3.13, ensuring full compatibility with newer environments. On Windows, tests have been stabilized by pinning the Numba version, and overall performance has improved thanks to memory leak fixes.

Additional workflow enhancements include a fix for GitHub Actions that previously failed when sending messages on closed pull requests, as well as the resolution of a staging workflow error. Multiple workflow and Docker tag updates were also made to improve deployment consistency.

Across the application, we’ve resolved a variety of bugs and refined core modules. This includes a fix to the SQL renderer for INTERVAL statements, updates to integration modules and Knowledge Bases, typo corrections, and improvements to static path checks. Unit tests for path traversal in files and BYOM were added, further improving security and reliability.

We’ve also added API metadata to the Config endpoint, improved file deletion handling through a dedicated API fix, and reduced the number of warnings triggered when MindsDB starts up.

Finally, we resolved an error with reranker limits, silenced the Pydantic model_ namespace warning, removed unused manual approval steps for forks, and replaced the old Starlette WSGI wrapper with a2wsgi for better performance and maintainability.

Documentation

We’ve updated all documentation and README files to clearly reflect the supported Python versions, ensuring users know which environments provide the best performance and reliability. The documentation now highlights compatibility for Python 3.10 through 3.13, helping developers set up MindsDB correctly from the start.

Across the docs, we’ve made numerous enhancements for better clarity and organization. These include updates to MindsDB object naming conventions, the Mintlify docs dependency, and improved guidance on configuration files, environment variables, a2wsgi configuration, GUI updates, and Bring Your Own Model (BYOM) usage.

In addition, we’ve refreshed visuals with an updated main image, fixed broken links, and added more detailed contribution guides to help the community get involved. Two new documentation folders have also been introduced — one focused on Hacktoberfest, and another featuring industry-specific use cases to showcase how MindsDB can be applied across different sectors.

SQL Operations

We’ve enhanced the SQL renderer to handle INTERVAL statements more efficiently and corrected an issue with LEFT OUTER JOIN syntax. These updates improve both performance and accuracy when running complex SQL queries within MindsDB.

In addition, we’ve refined how SQL skills are managed by Agents, ensuring smoother interactions when executing queries through AI-powered workflows. Connectivity and schema handling have also been improved — issues affecting Databricks and Oracle handlers that previously caused column and table lists to return incorrectly have now been resolved.

Users can also expect more efficient communication with MySQL APIs and smoother integrations with Snowflake and Oracle databases, resulting in faster, more stable data operations across the board.

OS GUI

This month’s releases brings a series of updates to improve the MindsDB interface, onboarding, and overall user experience.

Ollama has been added as a default model provider, available for use as both a reranking and embedding model. This addition gives users greater flexibility when configuring models directly within the MindsDB platform.

We’ve also fixed a bug where output results displayed in the wrong tab, and refreshed the sidebar and tab styles for a cleaner, more intuitive interface. In the SQL Editor, you can now stop a running process while executing queries — giving you more control and responsiveness when experimenting with data.

The web version has also been updated with improved error handling and clearer messages, making troubleshooting simpler and faster. To make learning easier, MindsDB’s documentation is now embedded directly in the GUI, so help and examples are always just a click away. 

We’ve refined the onboarding process to include a Terms & Conditions acceptance step, ensuring a smoother start for new users. Other improvements include fixing the tree-view scrolling bug in the data source dropdowns, enhancing case sensitivity for integration names in GUI forms, and adding comprehensive error handling to A2A agent streaming for better stability.

We included details about a new environment variable for managing PID File, which helps users handle process settings more easily. Additionally, we added a list of date-time functions and list commands to help users find what they need faster.

Conclusion

The October 2025 Product Updates mark another step forward in making MindsDB the most accessible, flexible, and production-ready AI platform for real-world data. With improved Knowledge Bases, expanded LLM integrations, a smoother SQL experience, and a more intuitive GUI, this update puts even more power and transparency in your hands.

Whether you’re running complex queries, experimenting with hybrid search, or managing enterprise-scale AI deployments, these updates ensure a faster, more stable, and more intelligent workflow from end to end.

Check out the full Release Notes here. Upgrade to MindsDB v25.10.1 today to take advantage of all the latest improvements, and stay tuned — more exciting features are on the way!