Skip to main content

v1.16 Explicit `/action` system, vision support

— PromptPanel

Update your image with:

docker pull promptpanel/promptpanel:1.16

and then re-create your container.

Features

  • Added vision support to Ollama / Google Gemini models.
  • Adopted the explicit /action system across all plugins.
    • /lookup for Document Lookup (RAG) plugin.
    • /append for appending a document to all base models plugins.

Enhancements

  • Summarize plugin returns one unified message now instead of multiple.
  • Fixed environment variable issue with Ollama.
  • Fixed document appending in Completion Model - added specific document_content context which can be used in Jinja templating.
  • Added Ollama call-out for when disabled.
  • Permanent fix for GPT-4o tokenizer, removed shim.
  • Improved thread titling prompt for models.
  • Improvements to Search window design.