awacke1/GraphicAINovel · GPT 4 0125 Preview Results

This set of runs have the version below:

MODEL DESCRIPTION CONTEXT WINDOW TRAINING DATA
gpt-4-0125-preview New GPT-4 Turbo
The latest GPT-4 model intended to reduce cases of “laziness” where the model doesn’t complete a task. Returns a maximum of 4,096 output tokens. Learn more. 128,000 tokens Up to Dec 2023

Observations:
GPT3.5 Turbo was generating full pass with three queries where each one was complete in 8 seconds.
4 1/25 preview - 60+ seconds. (around 8 times slower.

Output looks fundamentally different (colored markdown, different fonts, etc which were surprising.
Running two agents simultaneously leads to lag on both in fact first time testing had to close down simultaneous requests.