diff --git a/home/user/.config/pva/conf.d/13-ollama.conf b/home/user/.config/pva/conf.d/13-ollama.conf
index 1120c8d..735e743 100644
--- a/home/user/.config/pva/conf.d/13-ollama.conf
+++ b/home/user/.config/pva/conf.d/13-ollama.conf
@@ -12,8 +12,12 @@ text:"en","AIIDENTIFYIMAGE","What do you see on this image?"
 
 ai:"keyword","question"
 ai:"enable","false"
-ai:"port","11435"
+# we default now to ollama's normal port, not Alpaca anymore.
+ai:"port","11434"
 ai:"host","localhost"
+
+# You have to have this model installed, or it does not work out-of-box!!!
+
 ai:"model","llava:latest"
 ai:"mode","gapfiller"
 ai:"device","/dev/video1"
@@ -25,6 +29,15 @@ ai:"quality","70"
 ai:"languageprompt","antworte in deutsch."
 ai:"aliases","freisprechen=freetalk,schlüsselwort=keyword,lückenfüller=gapfiller,schlüsselwort modus=keyword,lückenfüller modus=gapfiller"
 
+# A note: Yes, we could readout ollamas list of installed models, but it wont work. the STT system will translate "llava:latest" to "laber doppelpunkt later ist" if you are lucky, so we need alias that are recognizeable.
+#         If you enter models you do not have installed or running and swap to them, "AI" stops working, but you can always switch back. 
+# Example:
+# ai:"models","normal=llava-llama3:latest,mond=moondream:1.8b-v2-fp16,lava=llava:latest"
+
+ai:"models","lava=llava:latest"
+
+command:"sprachmodell wechseln zu .*","AISWAPMODEL","",""
+command:"künstliche intelligenz wechsel zu .*","AISWAP","",""
 command:"was ist auf dem bild zu sehen","AIIDENTIFYIMAGE","",""
 command:"was ist auf den bildern zu sehen","AIIDENTIFYIMAGE","",""
 command:"was halte ich in die kamera","AIIDENTIFYCAM","",""