- Added support for vision models in WorkbenchModelSelector, including visual indicators.
- Introduced localized price tier labels for better user experience.
- Updated LLM presets to remove deprecated price tier labels and ensure proper model identification.
- Enhanced tests to cover new functionality, including image handling in multimodal contexts.
- Refactored e2e tests for model creation and editing to reflect UI changes and ensure accurate assertions.
- Improved unit tests for DOCX handling to maintain visual identity and shared media integrity.