Remote llama cpp not working with ollama setup

[eduardor2k]

What happened?

Hi, I'm running a llama cpp remote computer running Qwen3-coder-next, llama chat app works fine in port 8080, but when I connect ProxyAI, using ollama to that endpoint, model listing works but chats don't work, in theory it should work, ollama and llama cpp use openai api

I'm always getting empty replys

Relevant log output or stack trace

Steps to reproduce

1. setup remote llama cpp
2. select model
3. write chat
4. send

CodeGPT version

3.7.5-241.1

Operating System

Linux

[Soul-Burn]

Same thing on Windows.
ProxyAI 3.8.0-241.1
IntelliJ IDEA 2025.3.4

The data goes to the model, seen in the server logs, and computation happens. But not there's no result in the plugin.

It worked fine in IntelliJ 2023, older version of the plugin.