ChatGPT is a product. GPT-4o is a model. Claude is a product. Confused yet? This short guide maps the vocabulary we use across our directory and blog.
Headlines blur two words that mean different things: model and product. Mixing them up is not embarrassing—it is the default state of AI marketing. A quick map saves you hours of confusion when you read our Live Feed or AI News posts.
A model is the engine: weights, training data, and capabilities packaged as something researchers and developers can call through an API. Examples people name in the news include GPT-4 class systems, Claude-family models, Gemini variants, and open-weight releases you can download and run yourself. Models do not, by themselves, come with a polished chat window or a billing page.
A product is what you actually open in a browser or app. ChatGPT, Claude, Copilot, Perplexity, and Midjourney are products. They wrap one or more models with interface choices, safety filters, memory rules, plugins, and pricing. Two products can sit on similar models and still feel completely different to use.
Why does the distinction matter? Because announcements attach to one layer or the other. "We trained a new model" is a capability story—context length, reasoning benchmarks, license terms. "We shipped a feature in our app" is a product story—file uploads, team seats, enterprise controls. Regulators and enterprises care about both, but for different reasons.
When AI Nerd Network tags a Live Feed item or writes an AI News post, we try to name the layer correctly. If a lab blog says a model improved, we treat it as infrastructure news. If a company changes what paying customers can click, we treat it as a product shift. Getting that right is part of earning your trust.
Keep this glossary in your pocket while you browse. The firehose gets quieter the moment the vocabulary stops wobbling.
For breaking headlines across dozens of publishers, open our Live Feed. Current takes live on the Weekly Brief.