7.1 Model, Product, and Feature
Earlier chapters used “AI” as a convenient general term. When you choose a tool, however, many similar names appear. Think of a restaurant: a model is like the underlying capability of the cook, a product is the restaurant you enter, and a feature is a service such as ordering, delivery, or reservations.
A model is the underlying capability that processes input and produces output. A product is the complete application you can open, sign in to, and use. A feature is a particular tool inside that product, such as web search, file upload, voice conversation, or image generation.
Why can one name be confusing?
One product may let users select among several models, and one model may appear in several products or services. Free and paid versions of the same product may also differ in model access, features, usage limits, or file sizes. Saying “I use this company's AI” may therefore be incomplete.
A clearer description is: “I used this product, with this model or default mode, and enabled this feature.” ChatGPT, Claude, Gemini, Qwen, DeepSeek, Kimi, and Zhipu Qingyan are examples of product entry points familiar to everyday users. Model labels, modes, and features should always be checked in the product interface at the time of use.
When results differ, ask which layer changed. Was a different model selected? Was web search or file reading enabled? Did account access change? This is more useful than simply saying that “the AI suddenly became worse.”