Prompting
The same model can give a weak or a strong answer to the same question depending entirely on how you word the request. The prompt is the only steering wheel most users have.
A vague prompt leaves the model guessing what you want, so it rambles generically. Adding specifics, a role, a format, or a couple of examples narrows the space of good next tokens toward the response you actually need. This is not the model getting smarter; it is you removing ambiguity. Tap each style to see the output sharpen.
Dogs are domesticated animals kept as pets around the world. There are many breeds...
Check yourself
The model and its weights are identical across all four tabs. Why does the answer change so much?
Go deeper (free): Prompt Engineering Guide (free) ↗