I needed to do some LiteLLM hacky things to get some models working in Discourse AI.
Had you tried different reasoning models? Some LLMs only work with it enabled, and others with it on low (not minimal) or always default; that is usually heavy.
I needed to do some LiteLLM hacky things to get some models working in Discourse AI.
Had you tried different reasoning models? Some LLMs only work with it enabled, and others with it on low (not minimal) or always default; that is usually heavy.