new model: react-1, audio tags for TTS, generations renames, Microsoft SSO, and more

react-1 - A new model that lets you control performances in post-production. Upload your footage (filmed or AI-generated), then adjust the acting without reshoots. Select head, face, or lips and fine-tune delivery using the emotion wheel. Learn more about react models.
Audio tags in TTS - Add tags in square brackets like [excited] or [whisper] when generating text-to-speech to control emotion and delivery. Using at least one tag enables ElevenLabs v3 for more expressive speech.
Rename generations - Give your results meaningful names so they're easier to find after downloading. Click "..." on any history thumbnail and select "Rename". Defaults to project name + generation number.
Microsoft SSO - You can now sign in with your Microsoft account.

Improvements

Upload panel now auto-hides when all uploads complete instead of retaining previous state.
Smart suggestions now appear when generation results aren't great, with actionable steps to improve them.
Simplified model picker with settings moved to a separate dropdown in the same location.
Cloned voices now work over API. Copy Voice ID directly from the voices dropdown in Studio.

Bug fixes