[rstuart4133]

It's bait for an ad for EXXA, who apparently a mob who sells access to Llama 3.1 70B. It's pretty good bait. I found it interesting.

They say:

> Data processing with an LLM

After reading that I found myself wondering how badly it hallucinated. Unjustifiably, almost certainly, and to the extent it is justifiable it's probably less than the mistakes a human under time pressure to get content out the door would make. I guess I will have to get used to what is going to be the new normal.

[Blue_Cosma]

Hi, Constant here. I'm the Chief Mobster!

On hallucinations, larger models like Llama 3.1 70B still have a clear advantage, but they're also too costly to run at the moment on most API for those use cases. We built our API specifically for that reason.

I will try to make the disclaimer even more visible. We're not trying to hide anything here !