custom llm provider

Im running a custom LLM provider that uses the same API as openai.

I've set the env vars `OPENAI_API_HOST` and `OPENAI_API_KEY` and set my config as:
```
openai_params = {
model = "my-custom-model",
frequency_penalty = 0,
presence_penalty = 0,
max_tokens = 4095,
temperature = 0.5,
top_p = 0.5,
n = 1,
}
```

After that I can open the chatgpt window but when i issue a message the TUI just has the "thinking dots". Is there a way to get more debug info?

I have successfully use this API with other openai-compatible clients, so i think it might be a matter with playing with configs. but can't tell how to enable debugging for this plugin

Use the reported OPENAI_API_HOST, OPENAI_API_KEY, and openai_params configuration as the reproduction case. Trace the plugin's OpenAI-compatible request and response path, then identify how debugging is enabled and ensure request failures are surfaced instead of leaving the TUI on thinking dots.

Written by the indexing model from the issue text.