Support for open LLM models

The current implementation works great with using the openAI `gpt-4o` model. Alternatively, are there any open LLM models that can be used in replace of this? I tried with `Llama` and `Qwen` and both do not seem to do the job affectively especially with tool calls. We are looking to see how we can utilize graphchat with locally hosted models.

The issue names no files or tests. Start by tracing the current OpenAI gpt-4o integration and its tool-call handling, then review how model configuration is passed through the application. Done would mean identifying and implementing a defined path for locally hosted open models such as Llama or Qwen with reliable tool calls.

Written by the indexing model from the issue text.