Voxta

Voxta 151 Beta: In-Chat Diagnostics, Content Levels, and Key Fixes

Added 2025-08-30 18:09:08 +0000 UTC

Hello everyone,

Voxta 1.0.0-beta.151 is now available. This release focuses primarily on bug fixes and quality-of-life improvements, with a few new features added.

Key additions include the ability to open the diagnostics panel without leaving your chat, settings for different explicit content levels, and code highlighting for documents. This update also adds support for image generation through OpenRouter.

See the full list of changes below.

✨ Server Updates

Explicit Content Levels: You can now specify different explicit content levels for characters and scenarios (Prohibited, NotSpecified, Allowed, and Encouraged). The profanity filter will automatically enable for content set to "Prohibited."
OpenRouter Image Generation: Added support for image generation models available through OpenRouter.
Vision & Dependency Updates: Vision prompting has been improved, and the transformers library has been updated to version 4.56.
Key Fixes:
- Addressed an issue with F5-TTS voice trimming that caused repetitions and audio breaks, and fixed a missing dependency.
- Fixed a "get_logits_ith" error in Orpheus caused by a breaking change in llama.cpp.
- Resolved a Python integer conversion error for Sesame CSM.

⚙️ UI Improvements

In-Chat Diagnostics: The diagnostics panel can now be opened directly from the chat screen using either the left menu or the message's context menu.
Code Highlighting in Documents: The documents view now supports syntax highlighting for code blocks, making them easier to read.
Vision Playground Update: The UI has been updated, and you can now click on the image area to upload a new file.
Service Configuration: Added "Edit Config" buttons for all modules and services in the right-side panel. The UI now also shows which module is selected for a service, even when it's disabled.
Clearer Instructions: Improved the clarity of instructions for the ExLlamaV2 model syntax in the UI.

Links: