DBeaver Lite 25.3

December 8, 2025
Changes since 25.2:
  • AI Assistant: 
    • New engine support:
      • Anthropic Claude
      • Amazon Bedrock
    • AI Chat:
      • Added browser mode with lists, tables, and other complex formatting, enabled by default. The previous chat appearance can be chosen in Preferences.
      • Improved speech recognition: added sound visualization, pause detection, and the ability to adjust transcription settings.
      • Added the ability to cancel requests to AI.
      • Added a warning in AI chat when the conversation context is truncated.
      • Added new icons for AI tools and improved chat navigation: different conversation types are now marked with corresponding icons.
      • Improved error messages for OpenAI and Copilot.
    • AI functions (experimental): Added the ability to open the most popular DBeaver tools directly from the AI Chat. When this function is enabled, the assistant can trigger internal operations, such as exporting data, comparing schemas, creating connections, and opening editors. All functions are disabled by default and must be enabled individually.
    • Added the ability to automatically determine the proper context and send only relevant objects to AI (experimental): this allows you to use AI tokens more efficiently to reduce costs and improve the accuracy of results (using RAG technology). It is disabled by default. 
    • @ai command:
      • Fixed bug in AI command, which led to an error whenever a query contained a trailing delimiter.
      • Improved AI command handling for Oracle and other databases.
      • @ai command becomes disabled if the AI feature is turned off in the CLI or UI.
    • Configuration:
      • Added the confirmation for query types that DBeaver cannot recognize as valid SQL. 
      • Added the ability to test the connection to Amazon Bedrock, Anthropic Claude, Google Gemini, GitHub Copilot, and OpenAI engines.
      • OpenAI legacy API support was added.
      • Disabling AI now takes effect immediately without restarting the application.
      • Fixed an issue when the model name case was changed incorrectly.
      • Models that do not support temperature will now have the temperature field disabled.
      • Added the dropdown to the AI button on the main toolbar with the additional configuration buttons.
      • The option “Send table sample data” was moved to the Functions section. Now it’s called “Read table sample rows”.
  • SQL Editor:
    • Semantic analysis:
      • Added the ability to fix semantic errors using AI assistance.
      • Added support for temporary tables in semantic analysis.
      • Autocomplete proposals are no longer shown for values inside quotes. 
      • Fixed an issue when alias names were duplicated in the join table completion.
      • Fixed recognition of pseudo-columns in semantic analysis. 
      • Restricted the generation of aliases that match SQL keywords. 
    • Result grid minimizing now works correctly. 
    • Fixed an issue when the selected schema in new scripts resets on reconnect. 
    • Improved parsing and execution of multiple queries. 
    • Fixed an issue when the wrong ID was selected after a join operation. 
    • Resolved the issue with performance lag during long script scrolling. 
    • @include command processing was improved. 
    • Error processing in nested scripts was fixed.
    • `CTRL+Enter` shortcut was fixed in the Find/Replace panel.
  • Data Editor:
    • The panel context menu was redesigned.
    • Improved filter appearance and layout: unified background color, fixed separator line, and merged two panel toolbars into one.
    • Fixed dictionary lookup for composite foreign keys to show only distinct values in the Value panel.
    • Icons for data editing commands were updated.
    • Added the option to disable the formatting of numeric identifier columns.
    • The default shortcut for Advanced Copy with previous settings on macOS was changed to CTRL+SHIFT+4.
    • Removed the default shortcut for Select Row Count (Ctrl+Alt+Shift+C) due to a conflict with the Advanced Copy shortcut.
    • Data loading progress messages now show more details.
    • Fixed an issue when the boolean inline editor was read-only in tables without a primary key.
  • Metadata: 
    • Updated the UI for editing object comments.
    • Removed duplicated object name comments in the view DDL.
    • Redesigned panel with reduced empty space and improved field layout.
  • Navigator:
    • Added an option for case-insensitive alphabetical sorting.
    • Fixed an issue with table list reading in certain drivers (Databricks, Sybase, and others).
  • Data transfer:
    • Added the option “Ignore characters outside quotes” when importing from CSV.
    • Added BIGINT support and numeric type widening during the data import.
    • Input file selector UI now allows file name copy/edit.
    • “Write to the single file” option was fixed.
    • Added connection and database info to backup/restore wizards.
    • Fixed column casing issues when importing from XML.
  • Visual Query Builder: 
    • Added the ability to change the order of joins in the Join tab. The generated query is updated automatically based on the new order.
    • Added an option to hide all unused columns on the diagram. Only columns used in the query and key columns remain visible.
  • Connectivity:
    • Added support for connecting via JDBC URLs with custom protocol parts, such as “jdbc:mysql:loadbalance”.
    • Added a suggestion to reconnect after changing the driver version in the connection configuration.
    • Fixed an issue that prevented deleting network profiles.
    • Fixed an issue that prevented deleting an SSH jump server in connection settings. 
    • Network profile settings are now easier to find in Preferences.
  • Charts: 
    • Added an option to sort chart columns in chart settings.
    • Grid lines in all charts are hidden by default.
    • The result set refresh button now works for the chart.
    • Row limit in chart settings now works as expected.
  • Security:
    • Fixed an issue when the Master password was requested for setting the project password.
  • Miscellaneous:
    • Command line tool:
      • Added the ability to configure the “Show all databases” property via the Command Line.
      • Commands now return meaningful exit codes that indicate whether an operation succeeded or failed.
      • Opening of SQL files by using the command line of the file explorer was fixed.
    • UI:
      • All tabs for the same Production or Test connection are now color-coded for easier identification.
      • Fixed an issue when the connection color was incorrectly applied to result sets in text format.
      • Unified the appearance of toolbar separators.
      • Fixed broken colors in Dark theme after the Eclipse update.
      • Fixed truncated labels in Preferences on Linux.
    • The driver download window was redesigned.
    • Added an option to disable Windows Defender scanning at startup. Use the parameter windows.defender.startup.check.skip=true to skip this check.
    • Set NO_AT_BRIDGE=1 for Linux applications.
    • Fixed a startup issue in the Snap version related to the NO_AT_BRIDGE variable.
    • A potential security breach in the spatial data viewer was resolved.
    • Fixed a crash on Windows that occurred when the username contained non-ASCII characters.
    • Fixed an issue where notifications stole focus when the user was in another application.
    • Fixed an issue when the catalog selection was lost upon disconnect.
    • Improved panel and result set maximize behavior.
  • New driver: 
    • Amazon Aurora DSQL
    • Amazon DynamoDB
    • Greengage
    • FerretDB
    • Google Bigtable
    • Google Firestore
  • Driver updates:
    • Apache Derby was updated to 10.17.1.0
    • Amazon Athena was updated to 3.6.0 
    • Exasol was updated to version 25.2.3. It is no longer bundled with the application and must be downloaded manually.
    • Databricks was updated to 1.0.9-oss
    • Google BigQuery was updated to 42_1.6.3
    • Google Cloud Spanner was updated to 2.33.1
    • Redshift was updated to 2.2.0 
    • Snowflake was updated to 3.27.0 
    • SQL Server was updated to 13.2.1
    • H2GIS driver version was decreased to 2.2.1 due to a driver error
    • Yellowbrick now uses the native Yellowbrick driver 0.9.3
  • Databases:
    • Athena: Fixed identifier naming rules for the Athena dialect.
    • Clickhouse: 
      • Query canceling support was added.
      • Fixed several bugs related to incorrect type handling, including array, tuple presentation, and nested column presentation.
    • CUBRID: 
      • Fixed some issues with the full text query plan.
      • Fixed DDL generation for tables with multi-column keys.
      • Fixed an issue when a newly created table would disappear from the schema after a refresh.
    • Databricks: 
      • Improved OAuth User-to-Machine authentication and added support for Machine-to-Machine authentication.
      • Added host and URL selectors to the connection page
    • DuckDB: Fixed an issue when the catalog selection was lost upon disconnect.
    • Google BigQuery: 
      • Added support for Google Cloud OAuth authentication with refresh tokens.
      • Added the ability to create and delete Primary Keys via the user interface.
    • Ingres: Fixed driver download. 
    • Kingbase: Fixed unsupported vector type com.kingbase8.util.KBobject.
    • MariaDB: Fixed node costs on the Execution plan.
    • MySQL: Added “Compress” flag in the dump wizard.
    • Oracle: 
      • Privileges tab now opens for non-admin users.
      • Fixed an issue where user metadata information was not displayed unless ‘Always use DBA_ views’ was selected.
    • PostgreSQL: 
      • Spatial data types support was fixed.
      • Fixed filtering issues for composite types with identical column names.
      • Added support for hash indexes in the Metadata Editor.
    • Redshift:
      • Issues with network disconnects were resolved for all Redshift driver versions.
      • Fixed an issue that prevented the driver from updating.
      • Jobs and system databases are now hidden in the Navigator.
      • Fixed incorrect DDL generation for materialized views.
    • Salesforce: Added support for the Client Credentials Flow in OAuth authentication.
    • Snowflake: Procedures with multiple versions are now displayed correctly.
    • SQLite: 
      • Fixed an issue where foreign keys created with shorthand syntax were not recognized as references.
      • Fixed “Cannot determine target node for File” error.
    • Sybase/SAP ASE: Issue with the stored procedures body reading was resolved.
    • Teradata: 
      • Fixed an issue when HELP COLUMN ran on every column in a query, which caused significant lag.
      • The connection dialog was redesigned: some settings were moved to the Advanced tab.
  • Localization: 
    • German localization was improved.
    • Spanish localization was improved.