adding model_name as an env variable to change if we were to use gpt-turbo

[amrrs]

[Sentdex]

I've considered this change, but I haven't personally found GPT 3.5 to work reliably enough here to justify "suggesting" to people that it's even an option.

Have you been trying with GPT3.5 and disagree?

[achraf-oujjir]

For me, I am working with a free api key, so I am working with gpt-3.5-turbo and it is working well. I also had to change the sleep_time in the gpt_query method to 25. After every 3 commands, it had to wait more than 20 seconds to resend a new command.