Ask it questions about itself, like when its data cutoff is and what model it is.

It’d be really hilarious if it’s just chatGPT. “Local” llms are very unreliable about this, but sometimes they give you hints that they aren’t OpenAI.