Why change which LLM is used in the AI Assistant?

In Curiosity you can select which Large Language Model (LLM) is used in the app.

You can currently choose between:
ChatGPT 3.5 (OpenAI)
ChatGPT 4 (OpenAI, pro feature)
Phi 2B (on-device, needs 2.5GB disk space, beta)
Llama2 7B (on-device, needs 5GB disk space, beta)
Llama2 13B (on-device, needs 10GB disk space, beta)
Llama3 8B (on-device, needs 5GB disk space, beta)
Mistral 7B (on-device, needs 5GB disk space, beta)
MistralLite 7B (on-device, needs 5GB disk space, beta)

ChatGPT is a powerful online LLM system by OpenAI. Llama2 is an open-source LLM by Meta. The main differences are:
ChatGPT is a large, cloud hosted model. That means it may perform better in some cases. However, it requires some data to be sent to OpenAI. Learn more here.
Llama2 is run locally on your device. That means your data all stays local, but it means you have to download, store and run the model which requires a fast computer.

Curiosity also supports custom models for enterprises in the context of your knowledge. Contact us to learn more.

How to change which LLM is used in the AI Assistant

To change which LLM is used in the AI Assistant:
Go to Preferences (⚙️ icon in the left sidebar)
Click on AI Assistant in the left sidebar in Preferences
Select an LLM from the dropdown list (screenshot below)

Updated on: 22/04/2024