@Norm,

That error usually means ONLYOFFICE can’t communicate with Ollama, not that the model is broken.

Since the LLM works in the terminal, check that your API URL in ONLYOFFICE AI settings – make sure it points to http://SERVER-IP:11434 (not https and not 127.0.0.1 if Docker is used).

Next, Test Ollama API:

curl http://localhost:11434/api/tags

If this fails, the connection might be blocking in firewall and selinux.

sudo firewall-cmd --add-port=11434/tcp --permanent
sudo firewall-cmd --reload
sudo setenforce 0

If it works after this, SELinux is blocking it.

I used your article to do the same on Fedora 42 using a local llm and Ollama but when I query the chatbot I get “Something went wrong, please try reloading the conversation“.

LLM is running on command line fine, do you have any suggestions?