GPT-3 CLI open source, improving response time and cost.

We'll see if we can improve the response and reduce the per-command-cost with fine-tuning for text-davinci-002, the current prompt length is 840 tokens and the price is $0.02 for 1000 tokens, which is $0.017/command.