Many Meanings of the word Model.

19th March, 2026

Introduction

As a person who went through many phases, 3d graphic, real-time engines, programming and now AI, I noticed that there is a plethora of words that persist through these fields. At the very fundamental level, they might describe the same thing, but they are abstracted into a unique, niche fitting idea.

That's why, depending on your expertise you'll understand the word "model" differently. Mostly because you see it through a different abstraction or you use models in a conceptually different way. A model can be used as a predictive tool but it can also be used as a descriptor. Is some cases it's also prescriptive.

A 3D model of a model wearing wrapped newspapers with labels describing various models.
Created in Blender - a software that offers a comprehensive array of modeling tools to create, transform and edit models.

3D Model – my first model!

If you're a 3D Modeller working in Blender, 3Ds Max or other tool, you see a mesh that can be deformed, textured, animated or rendered. That all it was for me. Something I look at and manipulate with instant feedback.

At some point I got involved in interactive applications and worked alongside Unity (a real time engine) engineers. I realised “model” meant something completely else to them. They were asking about vertex orders and weights. Flipped normals, gizmos that weren’t matching Unity coordination system. (Damn you, right-hand coordinate system!). Model wasn’t a visual thing for them. It was a weird combo of constraints and very specific properties.

Then I talked to graphic programmers. The world I knew shattered to pieces. Literally. Those solid-looking shapes I'd been happily pushing and pulling for years? Not solid. Not even connected. Just points in 3D space, grouped into triangles, moving together while remaining fundamentally separate. The mesh - the thing I thought I understood completely - was an illusion the renderer was constructing 30+ times a second.

At this point there wasn’t even a model. It was trigonometry.

Programming

I went into programming iOS apps straight from working over a decade in 3D world. It’s absolutely hilarious at this point, how fast I run into a word “model” again. Specifically while trying to add persistent storage (Core Data in Apple world) into my app. The moment I saw “Data Model” I was like – what do you mean? There aren’t any vertices around to model the model!

To top it up, a Data Model wasn't the data neither. It held none. It was a set of rules describing something that didn't exist yet. Not a representation of a thing, but a contract about what a thing was allowed to be.

You know what the app was about? Water Temperatures. Yes. I was suddenly reading about forecasting models. A scientific predictive model.

The model I once knew as something to look at had quietly become something else - a contract, a description, a forecast. Same word. Completely different relationship to reality.

ML, AI, LLM

In 2022 ChatGPT hit the mainstream. Everyone is talking about Large Language Models. The word model got stretched further than I'd ever seen.

A model that dwarfs all the other models I’d encountered before.

A model I can.. talk to?!

I guess, at this point, looking for vertex to move around is in my DNA. So the first question I had was: Is prompting like vertex manipulation where all the maths is abstracted away but now, instead of a point hanging in a 3D space, I carefully craft strings to nudge an invisible probability distribution? Wait. Am I fetching a database? As usual my brief investigation unveiled yet another model. Specifically: EMBEDDING MODEL and vector databases that store these embeddings.

Machine Learning Model. AI Model. Large Language Model, Generative Model. An explosion of models. To organise my thoughts I tried to define each because obviously they are not the same.

ML Model ≠ AI Model ≠ Large Language Model ≠ Generative Model

ML Model	a learned function, inputs to outputs. You feed it data, it optimises until it finds the relationship. Linear regression is a model. A decision tree is a model. The maths is explicit.
AI Model	an umbrella term for anything imitating intelligent behaviour. A game enemy reacting to your movement is an AI model. Claude is an AI model.
LLM	specifically trained on language at scale. Its job is to predict what token comes next, over and over, on an almost incomprehensible amount of text. A technical term that quietly became a brand.
Generative Model	rather than classifying what something is, it learns what something looks like well enough to make new ones. LLMs are generative. So are image diffusion models, GANs, VAEs.

So what is a model?

What ties all of these together is relationships. A 3D model is a relationship of points in space. A data model is a relationship between fields and rules. An ML model is a relationship between inputs and outputs. Even a physical scale model is a relationship but of proportion.

A model is never the thing itself. It's the structure of how the thing holds together.

Every time I thought I understood the word, a new field showed me another layer underneath. And every layer, without fail, was describing relationships. Every time I looked underneath a model, I found another model.

At this point I think the universe is just some sort of model. 🤣