v2026.10.15
TL;DR
You can now watch your tests run live in the browser and take control, see every step and action on a redesigned timeline, set the effort level of Thunders AI, and use Squash TM to import tests, create tests, and sync results.
FEATURE
- Live browser: You can now see the browser running your tests in real time. Before, a run showed you only screenshots.
- Take control of a run: You can take control of the live browser when a step fails or at a breakpoint that you set. You can add the missing step and resume the execution.
- Record in the live browser: You can now turn what you do in the browser into steps. You need no separate recorder.
- Thunders AI effort levels: You now set the effort level of Thunders AI before you send a request. Use a low level for a quick edit and a high level for a long flow with many screens, conditions, and data.
- Redesigned test steps panel: The steps of a test now sit on a new timeline. Each step shows its state at a glance: successful, failed, interrupted, stopped, skipped, not run, or running.
- Squash TM: You can now connect to Squash from Thunders, import tests, create tests, and sync results with Squash TM.
- Authenticated profiles in Thunders AI: Thunders AI now offers to record an authenticated profile when it cannot sign in to your app.
- Display options across all lists in Thunders: Display options let you determine which information appears in your view.
- Public API: You can now stop a test run with an API token.
IMPROVEMENT
- Step types: A separate icon now shows the type of each step, such as AI, API, or conditional.
- Reusable Test Cases steps: A reused test case now shows its child steps in a tree under the parent step. The parent step shows the combined state of its children.
- Generate Test Cases in Test Plans: Generate Test Cases on a scenario now shows its cost in credits.
- Thunders AI for API apps: Thunders AI now stays in API mode for an API app and hides the browse suggestions.
- Thunders AI test sets: The chat of a test set now lists every test case that Thunders AI created.
- Thunders MCP: Thunders MCP now warns you when a step uses a variable that does not exist.
- File Compare and Visual Compare: Both steps now find a file through a variable. Visual Compare now accepts a baseline image of up to 40 megapixels.
- Browser slots: A run now waits for a free browser slot. Before, it failed after a few seconds.
- Saved selectors: Thunders now keeps its saved selectors when a run fails because the page did not load, tells apart repeated controls in a list by their row text and checks a saved selector again when it points to a different element.
FIX
- Dates in test steps: Steps that use dates now give reliable results.
- Page loading: A label that changes every second no longer holds up the page check. Thunders now ignores a stale disabled state when the page shows the control enabled.
- Thunders AI: A turn that fails after Thunders AI wrote output now stays as an interrupted turn that you can resume.
- Thunders MCP: Thunders MCP now lists only the browser locations that you can use.
- Element detection: Thunders now keeps the click target when a click step says "close" in its purpose.
- Test sets: A test set with no test cases attached now shows its runs.
- Test runs: A run now completes when only disabled steps remain.
- Date filters: Date filters in lists now work on a self-hosted PostgreSQL database.
- Test assets: Thunders now reads the Excel table of a template.
- UI polish: Thunders now returns you to the page that you asked for after you sign in. Thunders keeps the project section when you switch projects. The test run lists no longer show grid errors.
Breaking changes
- The API now stops at startup when
ASPNETCORE_ENVIRONMENTis missing or has an unknown name. Keep it set toSelfHosted, as the.env.examplefile says.
Migration steps
- Pull the new images and redeploy, starting with the API Service (it applies any database updates automatically).
upgrade urgency
Critical
Ready to ship faster with smarter testing?
