🐛 Unable to use GPT4all with default setup

[wouterdebruijn]

This PR fixes the issue of not being able to use the --reasoning_model=gpt4all and --parsing_model=gpt4all options without changing the source code. Mainly caused by two problems.

• The program exits with status code 1 when the OPENAI_KEY is not set.
• The default GPT4all model is outdated and no longer available through GPT4all

These issues have been fixed by the following methods:

• The force exit has been removed
• The default GPT4all model has been changed to mistral-7b-openorca.Q4_0.gguf

[nyck33]

Is that model outdated now?

(pentest-env) nyck33@lenovo-gtx1650:/mnt/d/infosec/pentestgpt$ pentestgpt --reasoning_model=gpt4all --parsing_model=gpt4all
Your CHATGPT_COOKIE is not set. Please set it in the environment variable.
Downloading: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 4.11G/4.11G [06:26<00:00, 10.6MiB/s]
warning: model 'mistral-7b-openorca.Q4_0.gguf' is out-of-date, please check for an updated version
warning: model 'mistral-7b-openorca.Q4_0.gguf' is out-of-date, please check for an updated version
warning: model 'mistral-7b-openorca.Q4_0.gguf' is out-of-date, please check for an updated version
Welcome to pentestGPT, an automated penetration testing parser empowered by GPT.
The settings are: 
 - parsing model: mistral-7b-openorca.Q4_0.gguf
 - reasoning model: mistral-7b-openorca.Q4_0.gguf
 - use API: True
 - log directory: logs
Do you want to continue from previous session? (y/n) y
Please select the previous session by its index (integer):
0. README.md
Please key in your option (integer): 0
You selected: README.md
Error when loading the previous session. The file name is not correct
Expecting value: line 1 column 1 (char 0)
New conversation.5987e8c0-f9a4-11ee-9e37-00155d791541 is created.

New conversation.6f94072a-f9a4-11ee-9e37-00155d791541 is created.

New conversation.8995ab1a-f9a4-11ee-9e37-00155d791541 is created.

- ChatGPT Sessions Initialized.
Please describe the penetration testing task in one line, including the target IP, task type, etc.
>

How can we update it again? Also the paper never mentions mistral so how's performance? There seem to be some models closer to gpt4? Like

[GreyDGL]

Please feel free to use your own models; PentestGPT does not rely on or is developed based on any of the open-source models. As for the paper, it was completed before the support to open-source models.