What Is a "Model"?

You've been chatting with "an AI," but you're really talking to a model — a specific AI system trained on data to generate responses.

"AI" is the general concept; a "model" is a particular version you can actually use. Common ones include GPT-4, Claude, Gemini, and Llama, each built by different companies.

The same prompt can produce very different results depending on the model.

What Are Parameters?

A model's size is measured in parameters — numbers it learned during training that shape how it responds.

More parameters generally means more capability, but also more computing power needed to run it.

GPT-4 has an estimated trillion+ parameters. On this site, we use Qwen3 0.6B — just 600 million. That's intentionally small.

Large models need expensive GPUs and lots of memory; a 0.6B model can run on a regular laptop.

The tradeoff is less power on complex tasks — but it's perfect for learning prompt engineering. If you can get good results from a small model, your prompts will work even better on larger ones.