Version 11.3.347
29 Dec 2025
Bugs have been squashed, performance has been boosted, and rough patches have been smoothed! VIDIZMO is running sleeker than ever.
What's New?
-
Audio Translation goes global! English still gets top accuracy, but now supports Spanish, Italian, Korean, Portuguese, Polish, Japanese, German, French, Arabic, Hindi, and more. Subtitles, anyone?
-
OCR Enhancements OCR now supports Perso-Arabic scripts! You can generate searchable and redaction ready content in Arabic, Urdu, and Sindhi. Your portal just got smarter with extended character support!
-
Embeddings Node Turn text into smart vectors for search & AI magic works with OpenAI, Azure, HuggingFace, Google, or your own API.
-
Graph Selection for Embeddings Now you can pick which graph to use when generating embeddings, choose from your portal catalog or let the system pick the default. Bonus: full audit logs included for peace of mind!
-
Mashup Node for Workflows CRUD your mashups and manage timed data right in your automated workflows!
-
JSON File Loader Node Pull in JSON files for OCR, embeddings, or AI analysis, workflow ready!
-
Text to VTT Converter Node Turn AI transcripts into neat WebVTT captions and attach them to your video files subtitles made easy!
Enhancements
- Custom Loop Labels in LLM Node No more vague “Looping” messages! LLM Node loops now show clear, descriptive labels right in the chatbot so users always know what’s happening.
Bugs
- The update option in Studio Space redaction settings is no longer mysteriously disabled.
- The default embedding generation graph no longer throws exceptions related to data operations.
- When you increase the workflow graph recursion limit, it now applies properly in test mode.
- No more stack traces printed in error notifications when clearing system variables.
- You can now run multiple approval nodes at the same time no more bottlenecks or “waiting your turn” moments!
- Subgraphs now run smoothly no more random errors or infinite loops!
- Chat sessions now get proper titles no more untitled mysteries!
- HuggingFace embedding generation now works flawlessly no more error hiccups!
- Failed expressions now raise clear exceptions so you’ll know exactly what went wrong!