Turbine On-Prem 26.4.1 Release

september 24, 2026 turbine 26 4 1 more flexibility, better control, smoother automation turbine 26 4 includes new capabilities and improvements that make automation easier to manage, integrate, and recover webhooks can now return customized confirmation messages, html landing pages, or json responses for automated integrations, while hero ai actions can return structured json that can be used directly by downstream playbook steps administrators also gain greater control over hero ai usage with monthly soft and hard credit caps and usage notifications to help monitor consumption in addition, playbook imports provide seamless event processing by allowing playbooks and webhooks to be enabled automatically after import and automatically reprocessing events that were skipped while the package was being imported together, these enhancements help teams run automation more predictably, with clearer integration outcomes, greater control over ai usage, and smoother recovery during content changes explore this release? take an interactive tour of the key features in turbine 26 4 with the guided walk through below user guide topics for this release using webhook triggers in swimlane turbine https //docs swimlane com/using webhook triggers in swimlane turbine webhooks https //docs swimlane com/webhooks hero ai native action https //docs swimlane com/hero ai native action hero ai credits https //docs swimlane com/hero ai credits hero ai models https //docs swimlane com/hero ai models import swimlane solution packages https //docs swimlane com/import swimlane solution packages organizing playbook in the canvas https //docs swimlane com/y61m organizing playbooks in the canvas drag and drop chart types https //docs swimlane com/chart types dashboard card operations https //docs swimlane com/dashboard card operations what's new in this release custom webhook responses and landing pages webhooks now support customizable responses for users and applications that invoke a webhook url an optional html landing page can be configured for browser users, a custom json response for api clients, and dynamic variables can be included from the incoming request responses can be previewed before saving, with html and json response bodies supporting up to 10,000 characters authentication is optional for html, json, and default responses, allowing html landing pages to be configured without webhook authentication authentication can be enabled when access to the webhook needs to be restricted existing webhooks continue to use their current behavior until a custom response is configured this capability is particularly useful for collaboration workflows and third party validation challenges that require a clear message or a specific response format for more information, see using webhook triggers in swimlane turbine https //docs swimlane com/using webhook triggers in swimlane turbine and webhooks https //docs swimlane com/webhooks structured json output from hero ai the hero ai native action can now return structured json that matches a schema you define select json as the output format and use the built in schema builder to define fields, mark required properties, and add constraints you can paste existing json to create a schema quickly, or discover output properties from a successful test the default none option remains available for standard text, html, or markdown structured output makes ai results more reliable for downstream playbook steps without extra cleanup in transform data or script actions when output format is json, mark as sensitive is not available on the result or on defined json properties other output formats still allow mark as sensitive when obfuscation is enabled note structured json output is currently available behind a feature flag contact swimlane support to enable it for more information, see hero ai native action https //docs swimlane com/hero ai native action hero ai credit caps and usage notifications hero ai credits now give account teams clearer monthly usage control and administrator notifications in addition to soft caps, accounts can use hard caps that stop hero ai automations when monthly usage reaches 100% of the allotment, instead of continuing into overages soft cap accounts continue to run according to their contract account administrators receive automated alerts at 75% and 100% usage through the in app notification center and email at 100% on a hard cap account, notifications state that ai automations may fail until credits are adjusted with swimlane cap type is managed with swimlane for the account contract the credits summary banner continues to show current usage alongside these threshold notifications note soft versus hard cap type is set on the account, typically by swimlane for your contract threshold usage notifications (75% / 100%) are currently available behind the hero ai credit notifications feature flag contact swimlane support to enable notifications for more information, see hero ai credits https //docs swimlane com/hero ai credits hero ai agents default to claude sonnet 5 hero ai agents now use claude sonnet 5 as the default model when custom llm is off, with claude sonnet 4 5 as fallback this applies to plan generation, remediation, playbook builder, component agents, and related agent traffic, including ai soc hero ai companion continues to use claude sonnet 4 5 if custom llm is enabled, keep the medium default on sonnet 4 5 when users need companion you cannot configure agent or companion routing in the product ui claude sonnet 5 is not available in the hero ai native action model list in turbine 26 4 for more information, see hero ai models https //docs swimlane com/hero ai models seamless event processing during playbook import events generated while a playbook is being imported or updated are automatically identified and captured once the import is complete, these events are automatically reprocessed using the newly imported playbook, helping ensure that no events are missed during the transition for more information, see import swimlane solution packages https //docs swimlane com/import swimlane solution packages improved playbook and component rendering for missing or inaccessible canvas references canvas now continues to render the playbook or component graph when a referenced component, connector action, webhook, or flow event trigger is missing or inaccessible unavailable nodes are displayed as placeholders with clear warnings, such as action not available or webhook not available, instead of preventing the playbook from opening broken input mappings are also identified as not available, while nodes with invalid configurations display warnings to help identify and correct the configuration transform data property references are validated consistently with other actions for more information, see organizing playbook in the canvas https //docs swimlane com/y61m organizing playbooks in the canvas drag and drop addressed issues canvas create variables dialog broke when inputs no longer matched the schema fixed an issue where create variables and update variable actions failed to render when inputs no longer matched the input schema those variables now appear as corrupted with a warning instead of breaking the dialog browser tab closed without confirming unsaved canvas work fixed an issue where closing the browser tab did not ask for confirmation while you were working in canvas turbine now prompts before the tab closes so unsaved work is less likely to be lost skipped disabled playbook events lacked a clear reason and retry path fixed an issue where events for a disabled playbook were hard to identify and recover those events now show a playbook was disabled skip reason, and you can retry them from operational health → events script action error properties were not available as component outputs fixed an issue where script action error properties could not be mapped to component outputs in canvas error details from script actions are now available as component outputs for downstream use property references in action outputs were incorrectly evaluated fixed an issue where property references embedded in action output values were evaluated again, which could cause unexpected recursive results output values that contain property reference text are now treated as data rather than re evaluated reports and dashboards black scrollbar rectangles appeared in reports in firefox fixed an issue where black rectangular artifacts appeared on report scroll areas in firefox reports now display correctly in current firefox versions scheduled report charts and tables were cut off when printed fixed an issue where vertical chart axes and table content were cut off in scheduled report output printed and scheduled report layouts now show chart and table content as expected horizontal bar chart titles were truncated in scheduled reports fixed an issue where horizontal bar chart titles were truncated in scheduled reports chart titles now display fully in scheduled report output pie and advanced pie chart labels were cropped — fixed an issue where pie and advanced pie chart labels were cropped and charts did not scale correctly on dashboards, in the report editor, and in scheduled reports labels and slice expansion now display correctly at supported tile sizes a 1x1 tile is not supported an advanced pie chart on a 2x2 tile can still cut off the total label size advanced pie charts to at least 3x3 so the total value and label remain visible for more information, see chart types https //docs swimlane com/chart types and dashboard card operations https //docs swimlane com/dashboard card operations scheduled dashboard pdf failed with relative date filters fixed an issue where downloading or sending a scheduled dashboard pdf failed when a relative date filter such as current month was applied scheduled dashboard pdfs now generate successfully with those filters scheduled dashboards rendered incorrect data or ignored filters fixed an issue where scheduled dashboard pdfs could show content from another dashboard, omit applied filters, or include a request error message when multiple schedules ran close together scheduled dashboard output now matches the intended dashboard and filters dashboard drill down did not match previous day report filters fixed an issue where previous day filters included the current day, so dashboard and report drill down results did not match yesterday and previous 1 day filters now return only completed prior calendar day(s) records and applications date/time picker was hidden behind the records listing filter panel fixed an issue where the date/time picker in the records listing condition builder was partially hidden behind the filter panel at normal browser zoom the picker now displays fully so you can select date and time values corrupted record restoration timed out on large datasets fixed an issue where restoring or counting corrupted records failed with a server error on large datasets those restoration requests now complete without timing out under the same conditions workflow conditions on reference fields did not run expected actions fixed an issue where workflow conditions that evaluated reference fields did not trigger the configured action as expected for reference field conditions, only set field value actions run after save on the backend; the ui clarifies this behavior svg logo upload failed for some svg dimension formats fixed an issue where uploading an svg as a co branding or tenant logo failed when the svg used certain dimension value separators logo uploads now succeed for those svg formats