--local should respect $OLLAMA_HOST rather than defaulting to localhost

[sammcj]

Is your feature request related to a problem? Please describe.

No response

Describe the solution you'd like

$OLLAMA_HOST is the standard way to define your Ollama server URL. It is used by various Ollama clients including the official Ollama command line tool.

At present --local uses a hardcoded api_base of http://localhost:11434

It would be nice if it defaulted to the $OLLAMA_HOST environment variable value (if found).

Describe alternatives you've considered

No response

Additional context

No response

[suizideFloat]

100% agree. Its very unusable when its restricted/limited to localhost. I guess a lot of people run ollama on a system with GPUs thats not their actual localhost. So please add that this works. I tried even with --api_base parameter but run into errors (llm.py and lightllm).

[mdlmarkham]

+10 to this one - in fact I run Ollama on 4 PCs... with the heavy models on the ones with better GPUs. It would be great if the client could understand this and route LLM traffic efficiently.