GPT-4 support & setting token limit

[bekher]

Great concept, it would be helpful if we could change the following via arguments:

1. Set OpenAI model to GPT4
2. Token limit to support bigger codebases and larger changesets