Update 1.5.2 GPT-4.5 Turbo Vision

OpenAI recently released new feature for gpt-4-turbo api for image recognition. Image sources are added to the chat context in base64 format and the model sees them every time it generates a response. This means that you can compare different images, clarify details and so on.

This feature can be used in many cases. For example, to use pictures instead of character descriptions:

Or to describe situations and environments:

Each picture takes 85 tokens in context. Unnecessary images can be deleted.

The model has 128,000 context size tokens.

Model in the list: gpt-4-vision-preview

Usage guide:

TavernAI 1.5.2 Node.js

https://github.com/TavernAI/TavernAI
TavernAI 1.5.2 exe

TavernAI 1.5.2 Colab