LLM information

Once you open the application, you will need to select a model to use. GPT4ALL provides you with several models, all of which will have their strengths and weaknesses. To know which model to download, here is a table showing their strengths and weaknesses.

Model	Size	Note	Parameters	Type	Quantization

Model	Size	Note	Parameters	Type	Quantization
Hermes	7.58 GB	Instruction Based Gives long responses Curated with 300,000 uncensored instructions Cannot be used commerciall	13 Billion	LLaMA	q4_0
GPT4All Falcon	3.78 GB	Fast responses Instruction based Licensed for commercial use	7 Billion	Falcon	q4_0
Groovy	8 GB	Fast responses -Creative responses Instruction based Licensed for commercial use	7 Billion	GPT-J	q4_0
ChatGPT-3.5 Turbo	Minimal	Requires personal API Will send your chats to OpenAI GPT4All is only used to communicate with OpenAI	?	GPT	NA
ChatGPT-4	Minimal	Requires personal API Will send your chats to OpenAI GPT4All is only used to communicate with OpenAI	?	GPT	NA
Snoozy	7.58 GB	Instruction based Slower than Groovy but with higher quality responses Cannot be used commercially	13 Billion	LLaMA	q4_0
MPT Chat	4.52 GB	Fast responses Chat based Cannot be used commercially	7 Billion	MPT	q4_0
Orca	3.53 GB	Instruction based Explains tuned datasets Orca Research Paper dataset construction approaches Licensed for commercial use	7 Billion	OpenLLaMA	q4_0
Vicuna	3.92 GB	Instruction based Cannot be used commercially	7 Billion	LLaMA	q4_2
Wizard	3.92 GB	Instruction based Cannot be used commercially	7 Billion	LLaMA	q4_2
Wizard Uncensored	7.58 GB	Instruction based Cannot be used commercially	13 Billion	LLaMA	q4_0