Hi there, did I understand correctly that I can only chat with one document at a time using the LLM function? The answer that the LLM generates to my question is therefore only based on the content of the one document and not, for example, on the contents of all documents of the created database?
@InAnYan This should be possible via the chat with groups, right?
Hi Philipp! Thank you for trying out the AI features!
You can chat with several documents. This is implemented via groups.
To do this, select a group in the left panel, click right mouse button, and then there will be a button “Chat with group”. After clicking on it, a new window will open and you can chat with several entries 
Thank you both. I would really like to try it. The documentation mentions that the AI funtionalities are built in/available in version 6. However, here JabRef download latest version I can download only version 5.15. Is this the version I need? If not - where do I get version 6?
AI features are not included in the 5.15 release version (or even the latest 5.16 release), but they are available in the development branch of JabRef.
To download development branch, visit the index - powered by h5ai v0.30.0 (https://larsjung.de/h5ai/), and choose the file that corresponds to your OS
Please create a backup of your library files (.bib) and your preferences before you start using the development version. There are some breaking changes.
JabRef 6.0 alpha will be released soon, but there are still some issues we want to fix before that. The documentation for it has already been released, because we have some power users that thankfully use the development versions of JabRef and provide excellent feedback.
@Philipp, note that searches and search-groups work differently in the development version, so you will definitely want to follow the advice to make a backup before upgrading. I needed to spend some time learning the new search syntax (Lucene) and adjusting groups that did not convert perfectly.
Remember that you can install the portable build of the development and/or released version if you want to have both available while trying the new functionality.