⚙️ Under the hood
Inference
Running a trained model: you give it input, it computes the output. Every single answer is inference.
💭 Imagine …
Training is university; inference is every working day afterwards.
🔎 Example
Every Copilot answer is an inference run in a data centre, token by token.
🎯 Why it matters
Inference costs compute, money and energy on every request. That's why pricing is per token and small models handle simple tasks.
🎓 Prefer step by step?
The AI Literacy course: 6 short modules with quizzes, a solid foundation in about 60 minutes.