1.11.0 Beta 1

Dyad v1.11.0 Beta 1 adds support for recording tests by interacting with the app preview, introduces Auto Sidekick for lower-cost implementation with frontier-model supervision, and improves model and reasoning effort selection.

Highlights

Record tests from the app preview

You can now record a test by starting a recording from the app preview and interacting with your app as a user would.

Dyad captures actions such as navigation, clicks, and form input, then uses AI to propose assertions for the completed workflow. The test proposal is added to the chat input so you can review, refine, or send it before adding the test to your app.

Recorded tests use more stable element selectors, and Dyad prevents conflicting app and Git operations while a recording is active.

Lower costs with Auto Sidekick

The new Auto Sidekick model combines frontier-model supervision with lower-cost implementation. GPT 5.6 Sol currently supervises the overall task while delegating straightforward, scoped work to GPT 5.6 Luna.

This can reduce AI costs while retaining intelligent coordination across the task. Delegated activity remains visible in the chat, and existing approval and tool controls continue to apply.

Improved model and effort selection

Model selection is now sticky for each chat. After you start a chat with a model, Dyad continues using that model even if you select a different model in another chat. This helps preserve consistency and makes it easier to remember which model each chat originally used.

Reasoning effort is now selected on a per-model basis from the model picker. Dyad remembers each model's effort preference and displays the current level alongside it, helping you manage cost, response speed, and reasoning depth.

Enhancements

• Queued prompts are now restored after restarting Dyad.
• Fixed an error that could occur when stopping a chat with queued prompts.
• The code editor file tree now marks files with unsaved or uncommitted changes.
• Commit dialogs can now open the diff for an individual file.
• Added manual redeployment for connected Supabase Edge Functions.
• Reload keyboard shortcuts now apply to the app preview without reloading Dyad itself.
• Improved questionnaire guidance when creating app blueprints.
• Fixed reasoning behavior and model-picker interactions for the Value model.
• App previews now block remote websites from embedding them in frames.

Feedback

If you run into issues or have suggestions, please help us improve by reporting a bug or uploading a chat session.