Hello everyone! This update mainly improves custom APIs, voice generation, and the overall Steam experience.
## Custom APIs
- Fixed an issue that prevented some custom LLM APIs from working because of browser cross-origin restrictions.
- Added and improved presets for DeepSeek, Google Gemini, Qwen, and other models.
- Improved compatibility with Qwen reasoning models.
- Improved connection tests and retry handling for custom models to reduce occasional loading failures.
## Voice
- StepFun voice now supports both the Standard API and Step Plan API.
- StepFun users can select a concurrency level that matches their account tier, helping reduce rate-limit errors.
- Fixed an issue where a Workshop character's reference voice could be previewed but failed to generate speech during a story.
- Improved support for reading and uploading reference audio in formats such as MP3 and WebM.
- Improved voice generation and voice matching across different languages.
- Added a voice indicator to the save list, making saves with generated voice easier to identify.
## Steam Experience
- The mute button in the top-right corner of a story now affects only the current story and no longer changes the global sound setting.
- Added an “Exit Game” button to the main menu.
- Simplified Steam error messages and removed technical error IDs that were difficult for players to understand.
- Fixed and improved the display and use of some Live2D and Workshop characters.
Thank you for continuing to report issues. If you still encounter problems after updating, please let us know through the Steam Community.