# What’s new

## What is Datalore On-Premises

Datalore On-Premises is a self-hosted version of Datalore, tailored for teams that want a collaborative data science platform set up on their own infrastructure.

To learn more about Datalore, watch the [Datalore overview](https://youtu.be/cn7I4TSCPoI) or [get in touch](https://www.jetbrains.com.cn/en-us/datalore/get-in-touch/) with our team.

The page chronologically lists the key features and updates of Datalore On-Premises releases. For more information, see the [Datalore On-Premises blog](https://blog.jetbrains.com/datalore/category/releases/).

## 2026

| Version | Fixes/Improvements |
| --- | --- |
| 2026.3 |      * Expanded the [Datalore agent skill](skills.html) beyond the notebook API to workspace-level features, with access granted through consent scopes. Workspace-scoped API tokens can now be created from workspace settings    * Added support for OAuth database connections in the [Query tab](queries.html)    * Enabled managing plan configuration from the admin panel, with changes applied without a server restart    * Added an instance-wide default for the number of [scheduled run](scheduling.html) results to keep    * Enabled running agents without privileged mode in Docker installations, keeping file mounts and reactive mode available    * Made the package manager search timeout configurable through `pip.conf`    * Preserved table cell state, including transformations, chart settings, and presentation mode, across page reloads in interactive reports    * Fixed Trino `write_dataframe` calls silently inserting no rows when the connection has no default catalog    * `automountServiceAccountToken` now defaults to `false` on Kubernetes agent instances and can be overridden in the settings    |
| 2026.2.4 |      * Fixed a high-severity vulnerability affecting public interactive reports    * Addressed an issue with refresh token support in session-only Trino OAuth    * Resolved HTML output iframe sizing after content collapse    |
| 2026.2.3 |      * Applied a security update in Hub    |
| 2026.2.2 |      * Enabled PKCE login support for agents to use with the Datalore notebook skill    * Fixed an issue where scheduled report updates didn't refresh downloadable files with isolated artifacts    |
| 2026.2.1 |      * Developed the [agent skill](skills.html) and API to work with Datalore notebooks    * Expanded API token configuration with additional expiration date options and scopes    * Activated LSP code insight by default in Datalore On-Premises    * Implemented workspace-level Kubernetes service account binding for cloud-native access configuration    * Added support for default credentials for Amazon Athena, Redshift, S3, Google BigQuery and Trino    * Added support for short-term AWS access keys for Amazon Athena    * Enhanced [Data explorer](data-explorer.html) with conditional formatting, chart drill-downs, plot history, and other improvements    * Launched [report collections](report-collections.html)    * Enabled [downloadable file attachments](interactive-control-cells.html#file-downloader) for static reports    * Introduced a [Query tab to the database connection dialog](queries.html) for running database queries outside the notebook environment    * Enabled exploration of connected database schemas in the workspace view    * Introduced low-disk-space warnings in notebooks and reports    * Fixed a critical security vulnerability > **Note:** > Details to be provided when the CVE is published    |
| 2026.1.1 |      * Switched to `fuse-overlayfs` for non-interactive runs to support newer kernels    * Bumped `gcsfuse` version    * Enhanced database connection stability    |
| 2026.1 |      * Added [instance-wide BYOK for AI Assistant](configure-ai-assistance.html#byok)    * Introduced [Data explorer cells](data-explorer.html) that allow you to filter, transform, and visualize your dataframes without writing additional code    * Introduced a [two-container setup](use-sidecar-container.html) for running the agent in non-privileged Kubernetes environments    * Added an option to restrict sharing outside a team    * Implemented a setting to limit the [number of scheduled run results to keep](scheduling.html#number-of-runs)    * Added search by name for databases, cloud storage connections, environment variables, and SSH keys    * Enabled [copying SSH keys between workspaces](ssh-keys-in-workspace.html#copy-ssh-key)    * Allowed [replicating notebook tabs in reports](build-a-report.html#replicate-tabs)    * Provided the option for [reports to use a different machine type](publish-a-report.html#machine-type) than the notebook    * Added errors from scheduled runs to the [issue center](issue-center.html)    * Introduced a setting that controls whether workspace files can be accessed from reports    * Added ability to include an `init.sh` script when configuring the environment in the Admin panel    * Provided public `execute_sql` methods that do not rely on existing SQL cells    * Fixed a medium-severity security issue related to session cookies    * Changed the environment variable format used to [include](configure-database-list.html#allow-databases) and [exclude](configure-database-list.html#hide-databases) available database connector types > **Note:** > Configurations created for earlier versions require updates    |

## 2025

| Version | Fixes/Improvements |
| --- | --- |
| 2025.6.2 |      * Applied a security update in Hub.    |
| 2025.6.1 |      * Fixed an issue in the Kubernetes installation that prevented agent machines from being correctly tagged    * Allowed configuring the internal database connection pool using the [DB_MAX_CONNECTION_POOL_SIZE environment variable](server-configuration.html#db-max-connection-pool-size).    |
| 2025.6 |      * Launched the [Issue center](issue-center.html) to make it easier to monitor and manage errors in published reports    * Implemented the [Usages tab](databases-as-workspace-resources.html#find-usage) to display all notebooks and SQL cells where a database connection is used    * Added the [All reports page](datalore-concepts.html#reports)    * Allowed [downgrading users to viewers](manage-users.html#user-to-viewer) and improved the viewer UI (Datalore On-Premises only)    * Enabled new DuckDB-based tables by default and added an option to switch to legacy tables in the Admin panel    * Enabled sharing single reports from shared workspaces    * Added an option to automatically recalculate reports upon opening without prompting the user    * Added [support for Anthropic Claude Sonnet 4.5](configure-ai-assistance.html#select-model) (Datalore On-Premises only)    * Added automatic environment installation for pip environments on machine start when the environment and Python version are specified in the Admin panel, powered by uv    * Included links to the results of scheduled runs in email notifications    * Added support for background computations in reports (Datalore On-Premises only)    * Implemented a method to [force-stop machines](manage-running-machines.html#force-stop) stuck during graceful shutdown    * Upgraded the Recents section to include both notebooks and reports    * Fixed a UI issue preventing users from logging in to databases    |
| 2025.5.1 |      * Added a setting that controls whether users are prompted to recalculate interactive reports when opened    * Implemented multiple client-side fixes for Viewer users and notebook editor behavior in read-only mode    |
| 2025.5 |      * Introduced [workspace backups to Git repositories](git-backup.html)    * Interactive reports can now be [shared with currently selected parameters](interactive-reports.html#share-report-with-parameters)    * Added mobile layout for reports to improve usability on smaller screens    * Introduced multiple improvements to interactive tables. Table data can now be [downloaded as .xlsx files](interactive-table-output.html#download-table) and [copied in the TSV format](interactive-table-output.html#copy-table)    * Added a Data visualization Tool that allows building more complex charts    * Added the ability to [select the default model for AI assistance](configure-ai-assistance.html#select-model). In Datalore Cloud, switched the AI assistant to a newer model    * Supported OAuth2 authentication for Trino    |
| 2025.4.1 |      * Introduced multiple security improvements and addressed CSRF- and XSS-related vulnerabilities    * Resolved a rare bug when interactive tables were not displayed    |
| 2025.4 |      * Added [integration with Google Drive](connect-to-google-drive.html)    * Introduced support for multiple Python versions in conda environments    * Changed the default Python versions in conda environments to 3.11    * Resolved issues affecting Plotly chart rendering and improved application performance across multiple areas    |
| 2025.3.1 |      * The baseline Python interpreter has been updated to Python 3.11    * Consecutive cells in notebooks [can be folded](datalore-notebook-structure.html#collapsible-sections)    * Notebooks cloned within the same workspace preserve the original environment and context    * Reports can be [organized in folders](manage-reports-in-workspaces.html#organize-reports-with-folders)    * Text fields support [default values](interactive-control-cells.html#text-control-default-value)    * Introduced [workspaces import from .zip files](import-workspace-from-zip.html)    |
| 2025.2 |    > **Warning:** > Deprecation notice: Starting from Datalore On-Premises 2025.3, the baseline Python interpreter will be updated to Python 3.11. > > > > No further updates are planned neither for the previous baseline images with Python 3.8 nor for any of the previously custom released images (Python 3.10). > > > > If you need to have a customized interpreter version, consider [building the custom agent image.](customize-or-update-environment.html)       * [Snowflake](snowflake.html) keypair authentication is now supported    * Custom JDBC driver [can also be used for the Datalore server itself](https://www.jetbrains.com.cn/en-us/help/datalore/2025.2/using-custom-postgres-driver.html), not only [for the data sources.](custom-drivers.html) > **Note:** > Custom JDBC drivers are only supported in Datalore On-Premises 2025.2.    * Introduced [tabs support](build-a-report.html#prepare-a-report) in Report Builder    * DuckDB is available as datasource for SQL cells    |
| 2025.1 |      * Deprecated features removed: notebook-level Git repositories, account-level SSH keys and secrets    * Data: SSH tunnel support for new SQL sessions    * Home page: welcome screen for new users    * Git integration: creating workspaces by importing Git repositories    * Admin panel: new appearance and improved user experience    |

## 2024

| Version | Fixes/Improvements |
| --- | --- |
| 2024.5.1 | A minor release that:     * Fixes output size for Ub requests    * Prevents automatic deletion of Hub users in Datalore on force-sync    |
| 2024.5 |      * Data: * SQLite database support * Attached data tool: reworked Attached data tool    * Interactive controls: value lists of dependent dropdowns automatically updated based on the parent dropdown change    * Interactive tables: * Not equal filter added * Find and Replace option added    * SQL cells: new database connection method to improve SQL cell performance    * Export: improved report export to PDF    * Sharing: Manage type added for granting full access when sharing workspaces and notebooks    * Editor: opening files from stacktrace and Go to declaration enabled    * Reports: copy output function introduced    * Instance administration: cloud storage providers added to instance-wide data connectors    * Hub integration: automatic user email update after being changed in Hub    * Computation management: external agents support    * Audit: simple option to configure logging level for agents    * Security fixes for: notebook import, banned user notifications, and XSS-related vulnerabilities    |
| 2024.4.1 | A security fix that handles XSS vulnerabilities |
| 2024.4 |      * SQL cells: option to limit returned rows to 500    * Enhanced folder uploading procedure    * Dropdown control: searchable list of values, Select all and Clear oll options    * Editor UI improvements: better cell visibility, clearer cell status indication, more accessible layout of cell toolbar, and more    * Quick notebook access from the editor    * Polars DataFrame support    * Python `input()` commands now supported for interactive reports    * Fixed issue of pages not loading for Datalore set up with an HTTP proxy on a non-standard port    * Fixed issue of database schemas not refreshable for users logged in via OAuth    * Bundled PostgreSQL server updated to Postgres 15    |
| 2024.3 |      * Scheduled runs: advanced run data handling    * Smoother environment setup for R and Scala    * Early Terminal access during setup    * Support for ipydatagrid versions 1.3.0 and 1.3.1.    * Customizable Git pull size limit    * Correlation chart fixed for Visualize tab    * Fixed rendering of ipywidgets outputs in the report builder and reports    * Markdown cell variables now updated in scheduled reports    |
| 2024.2.3 | SQL cell output: fixed handling of DataFrames containing long strings |
| 2024.2.2 |      * VM options support in database connection dialog    * Improved performance of notebook scheduler    * Fixed workspace menu display when viewed in Admin mode    |
| 2024.2.1 |  Home page:     * Improved usability of notebooks files view    * Fixed folder, trash, and report search    * Fixed issue of specific workspace names    * Fixed URLs inside Data section    |
| 2024.2 |      * New Home page: * Links to respective reports in the Notebooks list * All workspaces listed together and easily accessible * Data section with quick access to all data sources in the workspace * Option to enable workspace-level management of SSH keys and Git libraries    * Editor: * Enhanced Ask AI * AI-powered Autopilot used to get code suggestions based on a custom notebook objective or notebook content analysis * Support for variables in Markdown cells * Code syntax highlighting for Markdown cells    * Data: * Windows native authentication for Microsoft SQL Server databases * Azure OAuth2 authentication for Snowflake databases    * Reports: option to restrict downloading or editing copies of report notebooks    * Administration: * Customizable full storage alerts * Custom PyPi server configuration support * Customizable notebook code import limit    |
| 2024.1 |      * Editor: * Download file interactive control added * Option to load HTML outputs form a separate domain for security and performance purposes * Optimized performance for large notebooks * More edit options for interactive table outputs * Vim text editor preinstalled for Terminal    * Data: * configurable mounts for Docker agents * Instance wide data connections created and managed by admins    * Collaboration: keeping report calculation results within the user session    * Administration: * Option to add custom time intervals to background computation settings * Environment tab for advanced environment configuration * Group plans option for user group management    |

## 2023

| Version | Fixes/Improvements |
| --- | --- |
| 2023.6 |      * Editor: * New UI design (tool icons and menu layout) * Sticky Table of Contents with active section highlighting * Soft-wrapping for text outputs    * Computation: Notebook ID and notebook owner ID as environment variables    * Administration: * Dashboard tab for status monitoring * Limitation for viewer type users dropped    * Data visualization: lets-plot library support for air-gapped environments    * SQL cell: proactively starting sessions for each data source when opening the report    * Database connections: ability to import connection parameters in XML format, Okta OAuth authentication for Snowflake    * Scheduled runs: starting agents without using the agent pool    |
| 2023.5.1 | A security fix that handles XSS vulnerabilities |
| 2023.5 |      * Editor: improved interface for opening recent notebooks from the editor main menu    * Data: * Custom JDBC drivers support * Snowflake and BigQuery OAuth authentication method * Fixed issue of multiple Git repos with different authentication keys    * SQL cells: * Table name parametrisation * Proactive start of SQL sessions to ensure faster execution of SQL cells    * Collaboration: mentions enabled for comments in reports and notebooks    * Report builder: dynamic height enabled for more than one cell in a row    * Admin panel: note field for profile deletion action    * Interactive reports: notifying users of recalculations without time delay    * Home page: option to empty all trash    |
| 2023.4 |      * Comments feature for notebook and report collaboration    * SQL cell: option to query DataFrames    * Export to database cell    * Databases: cloning data sources between workspaces    * New data source type: SMB/CIFS folders support    * Collaboration: Viewer user role, report sharing by email invitations, STARTTLS support for sending emails    * Interactive cells: file uploader control, improved usability of dropdowns    * Admin panel: improved user list view, background computation management on user level    * Scheduling: parameterized runs    * Cells: full width mode for cells, improved display of cell elements/fonts    * Reports: fixed behaviour of dependent interactive controls    * Report builder: cells multiselect and multidrag, UX improvements    |
| 2023.3.1 | A security fix that resolves session authentication vulnerabilities and improves session cookie management  |
| 2023.3 |      * Data: Vault AppRole authentication method for Athena and other database types    * Scheduling: multiple schedules for one notebook    * Report builder: layout improvements, more flexible drag and drop    * Fixed variable selection in the dropdown interaction control settings    * Shared notebooks cloned now together with attached files    * Improved file saving when executing scheduled runs    * Fixed value setting for the slider interactive control    |
| 2023.2 |      * Report Builder: Undo/Redo support    * Search for notebook content option    * Environment: * Secrets deprecated, replaced by Environment variables * Native R packages support    * Fixed agents recovery after plans disablement    * Fixed issue of ToC not displayed properly after configuring Datalore to a non-default port    * Improved Scala kernel performance    * DataFrames download issue fixed for Firefox    * Swipe left gesture disabled in the editor    * Admins allowed to access all Datalore workspaces    * IPywidgets rendering issues fixed    |
| 2023.1 |      * Support for several API requests    * Computation: 1. Fixed issue of stopping computations for scheduled runs 2. Fixed machine performance in case of Internet outage 3. Reactive kernel improvements    * Data sources: 1. Bucket content now visible to viewers 2. Ability to create cloud storage data sources on workspace level    * Improved report builder performance    * Ability to download workspace content as .zip files    * Fixed procedure of public link revoking    * Fixed metric cell issues    * Fixed issue of ipywidget rendering    * Improved procedure of data source connection checking    * Fixed issue of Terminal unresponsiveness after system sleep    |

## 2022.3

| Version | Fixes/Improvements |
| --- | --- |
|  | Release updates |  |
| 2022.3.1 |      * Data output: Table and Raw tabs improved, table viewing and editing issues fixed    * Auditing: improved server data logs    * Statistics tab: support for Kotlin, Scala, and R fixed    * Code execution: fixed cell linking in output after re-evaluation    * Reports: fixed display of scheduled runs    * Scheduling: reports not affected by stopping associated scheduled runs    *     * Computation management: fixed scheduled runs started in the background and concluded after re-opening the notebook    |
|  | Major release details |
| 2022.3 |      * Sharing and collaboration: designing and publishing reports with the Report builder tool    * Editor: 1. Interactive table output 2. Metric cells    * Logging: 1. Head dump and thread dump download options 2. Server logs accessible from the Admin panel    * User management: 1. Group management interface (including Hub-attached groups) 2. Option to disable user registration 3. Option to create user accounts from the Admin panel 4. Contact system support (search and auto-completion)    * New landing page    * Interactivity controls: date picker control added    * Database integration: 1. SSH tunneling 2. BigQuery database connection support 3. Improved S3 bucket integration interface    * Background computation: default option support    * Administration: advanced user activity logging    |

## 2022.2

Release updates
: | Minor version | Fixed/Improved |
: | --- | --- |
: | 2022.2.3 | New licensing: free license for groups of 1-4 users (not available now)  |
: | 2022.2.2 |      * Databases: URL field input limit extended in database connection interface    * Collaboration: interactive reports made available to unauthorized users    |
: | 2022.2.1 |      * Login after adding Hub enabled    * One-by-one execution of multiple SQL cells on the Run all action    * 5-minute limit for SQL query execution removed    * Ability to stop database querying by explicitly stopping SQL cell calculation    * Presto database support via SSL    * Default database schemas displayed in detailed view    * Ability to fix database credentials inside the notebook without reattaching    * Scheduled runs fixed    * Improved S3 datasource interface to support custom options    |

Major release details
: | New features |      * New installation option: possible to install Datalore On-Premises using easy-to-set-up Docker-based configuration    * Coding and execution: 1. Scheduled runs for notebooks 2. Parameterized SQL statements 3. Computation tab    * Collaboration: collaborative editing of attached files    * Databases: 1. Limiting database schemas for introspection 2. MS SQL Server support    |
: | Improvements |      * Meaningful names of files downloaded from interactive reports    * Database passwords no longer exposed in Test Connection pod    * Collaborators allowed to change machine type in shared notebooks    * More cut-off timer options for background computation    |

## 2022.1

Release updates
: | Minor version | Fixed/Improved |
: | --- | --- |
: | 2022.1.4 |      * Added instance and storage gift codes    * Jedi disabled by default to speed up kernel completion    * Improved kernel stability following non-critical code errors    * Fixed opening of notebooks with erroneous syntax info    * Fixed dropdown value update in interactive reports    * Improved kernel stability when using interactive controls in Reactive mode    |
: | 2022.1.3 |      * Improved overall performance, especially when working with big DataFrames and using Statistics tab or Variable viewer    * Fixed analysis server    |
: | 2022.1.2 |      * Restarting kernel when detecting out of memory status to ensure uninterrupted cell execution    * Faster DataFrame rendering    * Data output download allowed to unauthorized users    |
: | 2022.1.1 | Fixed Spring Core RCE vulnerability  |

Major release details
: | New features |      * Statistics tab: used to provide in-depth information on DataFrames    * Table viewer: editor component for reviewing and editing CSV and TSV files from Attached data    * Visibility control: cell type for collapsing other cells in interactive reports    * Сut-off timer for background computation: option for setting the machine to shut down after a specific period of inactivity    * User activity logging: feature used to track cell execution by a specific user    * Reactive mode in interactive reports: automatic recalculation following user interaction with the report    |
: | Improvements |      * Enhanced reports: 1. Full width mode 2. Automatically hidden empty code cells 3. Checkbox control (for interactive reports only)    * Variable viewer: providing information on notebook object sizes    * Notebook structure: ability to use duplicated worksheets    * Improved editor experience: 1. Viewport extended when scrolling past the notebook last cell 2. Better indication of cells queued for processing    |

## 2021.3

Release updates
: | Minor version | Fixed/Improved |
: | --- | --- |
: | 2021.3.5 |      * Hub updated to 2021.1.14194    * Database connection sharing options: Shared workspace * Editors can create and edit database connections, and attach them to the notebooks of the shared workspace. * Editors can create and execute SQL cells for any connected database. * Viewers can view database connection lists and respective database schemas. Home workspace notebook Editors can create and execute SQL cells for any connected database.    |
: | 2021.3.4 |      * Hiding worksheet inputs and outputs    * Improved matrix rendering for the R kernel    * Latest Lets-Plot library version    * Visualize tab: 1. Bigger charts, improved chart settings layout 2. One-click chart export to a chart cell 3. R kernel support    |
: | 2021.3.3 | log4j-related security issue in Hub |
: | 2021.3.2 |      * Rendering of Pandas series with custom indexing    * Reactive mode for Kotlin    |
: | 2021.3.1 | log4j-related security issue |

Major release details
: | New features |      * Database connection: interface for connecting to databases from the Home page and the editor    * Native SQL cells: cell type used to retrieve data from connected databases without passing user's credentials to the environment    * Interactive reports: feature for publishing notebooks as pages that collaborators can dynamically change using interactivity elements    * Chart cells: cell type used to build customizable visualizations based on DataFrames in your notebooks    * Interactive controls: dropdown, slider, and text input elements to quickly customize output without manually changing the code    * Cell visibility control: interface used to select what input and output to hide from viewers    * More language support: R and Scala available as notebook languages    * Reactive mode: top-down evaluation with automatic recalculation of cells below the modified one    |
: | Improvements |      * Environment: customization via initialization scripts in the `init.sh` file    * Expanded Python support: Python 3.6 and later versions available with respective code insight features for each    * Visualize tab: more plot types added (point, bar, area, correlation)    * More bucket types: Google Cloud Storage buckets attachable to notebooks    * Last cell execution info: last execution time shown when hovering over a cell    * Cell collapsing: ability to collapse specific cell inputs and outputs for better coding experience    |