📖 Quick glossary

• Prompt: the instruction message you give the AI.
• Zero-shot: a prompt with no examples.
• Few-shot: a prompt with 2-3 “input → output” examples to fix tone and format.
• Chain-of-Thought (CoT): having the AI detail its reasoning before the answer.
• Reasoning model: a model that “thinks” internally before answering (CoT built in).
• Tree-of-Thoughts (ToT): exploring several idea branches, evaluating them, then choosing.
• Graph-of-Thoughts (GoT): combining/merging ideas from different branches (advanced).
• Tags: markers (e.g. <text>…</text>) separating instructions and content.
• Portable prompt: a well-structured prompt that works on any assistant without rewriting.
• Multimodal: the ability to process and/or produce several formats (text, image, voice, video).
• Context window (context window): how much text a model can “keep in mind” at once. The larger it is, the more documents you can feed it.
• Fast mode / reasoning mode: two variants of the same assistant — one answers fast (standard model), the other “thinks” before answering (reasoning model).