I truly dislike them (and Mistral and others) purposely not comparing it against GPT3.5 or 4, of which a vast number of developers are familiar with.

It makes me thing it's nowhere near in terms of performance.