Lesson 1: Platforms vs Models — Why ChatGPT Isn't "One AI"

Course: Choosing the Right AI Model (Personal Pathway, Free) Estimated reading time: 8 minutes Last updated: 2026-02-19

---

The Biggest Misconception in AI Right Now

If someone asked you "What AI do you use?", you'd probably say "ChatGPT." Fair enough — it's the one everyone knows. But here's the thing: ChatGPT isn't an AI model. It's a platform. And behind that platform are several very different AI models, each with different strengths, speeds, and price tags.

This distinction matters more than most people realise. It's the difference between saying "I drive a Toyota" and "I drive a car." Knowing the specific model changes everything about what you can expect.

What's a Platform? What's a Model?

Let's make this concrete.

A platform is the website or app you interact with. ChatGPT, Claude, Gemini, Copilot — these are platforms. They're the interface: the chat window, the buttons, the settings. Think of a platform like a restaurant. You walk in, sit down, and order from the menu.

A model is the AI engine running behind that platform. GPT-4o, Claude Sonnet 4.6, Gemini 1.5 Pro — these are models. They're the chef in the kitchen. Different chefs have different specialities, different speeds, and very different price points.

Here's where it gets interesting: a single platform often gives you access to multiple models. When you use ChatGPT, you might be talking to GPT-4o one moment and GPT-4o mini the next, depending on your subscription and what you're doing. The platform looks the same. The model behind it has changed.

Why Does This Matter to You?

Because not all models are created equal — and the one your platform defaults to might not be the best choice for what you're trying to do.

Let's say you want to write a cover letter. A fast, cheaper model might do a decent job. But if you need to analyse a 50-page contract, you might want a model with a larger "context window" — the amount of text it can hold in its head at once. If you're writing code, some models are significantly better than others at understanding programming languages.

When you don't know models exist, you're stuck with whatever the platform gives you by default. When you do know, you can make better choices.

A Quick Tour of How This Works in Practice

Let's look at ChatGPT specifically, since it's the one most people know.

When you use the free version of ChatGPT, you're typically interacting with GPT-4o mini — a smaller, faster, cheaper model. It's perfectly capable for everyday questions and quick tasks.

If you pay for ChatGPT Plus (US$20/month), you get access to more powerful models like GPT-4o and the reasoning-focused o1 and o3 models. Same platform, different engines under the bonnet.

OpenAI (the company behind ChatGPT) also offers these models through an API — a way for developers and businesses to plug the AI directly into their own software. Same models, completely different interface.

This pattern repeats everywhere:

• Anthropic makes Claude models (Haiku, Sonnet, Opus) → accessed through the Claude platform
• Google makes Gemini models (Flash, Pro, Ultra) → accessed through the Gemini platform
• Microsoft uses various models (including OpenAI's) → accessed through Copilot

The Restaurant Analogy, Extended

Think of it this way:

Some restaurants have one chef. Some have several, and you can request which one cooks for you. That's exactly how AI platforms work — and knowing this gives you a choice you didn't know you had.

The Model Names Can Be Confusing — That's Normal

AI model naming is, frankly, a mess. Here's a quick decoder:

OpenAI models (used in ChatGPT):

• GPT-4o — their main workhorse ("o" stands for "omni" — it handles text, images, and audio)
• GPT-4o mini — smaller, faster, cheaper version
• o1, o3 — "reasoning" models that think step-by-step before answering (slower but better at complex problems)
• GPT-5.4 — released March 2026, representing continued rapid iteration from OpenAI with significant capability improvements

Anthropic models (used in Claude):

• Claude Haiku 4.5 — fast and affordable
• Claude Sonnet 4.6 — balanced (the one most people use)
• Claude Opus 4.6 — most capable

Google models (used in Gemini):

• Gemini 2.0 Flash — fast and efficient
• Gemini 2.0 Pro — more capable

March 2026 context: The pace of new model releases has accelerated. Multiple providers released significant updates in early 2026 alone, including the GPT-5.4 release in March. This rapid iteration means model capabilities you rely on today may be significantly different by year-end.

Don't worry about memorising these. They change regularly. The important thing is understanding the concept: every platform has multiple models, and you often have a choice.

What You Should Do Right Now

Next time you open ChatGPT, Claude, or any AI platform, look for a model selector. It's usually a dropdown menu near the chat input. Click it. See what options you have.

You might be surprised to find you've been using the default when a better option was available — or that you've been paying for capabilities you don't need.

That awareness alone puts you ahead of most people using AI today.

---

Key Takeaways

• ChatGPT is a platform, not a model. The AI engine behind it (GPT-4o, o1, etc.) is the model.
• One platform, multiple models. Most AI platforms let you choose between different models with different strengths.
• The default isn't always the best choice. Knowing what models are available lets you pick the right tool for the job.
• Model names change frequently. Focus on understanding the concept (tiers of speed, capability, and cost) rather than memorising specific names.
• You probably already have more options than you realise. Check the model selector in your current AI platform.

---

Practical Exercise

Time needed: 10-15 minutes

1. Open the AI platform you use most often (ChatGPT, Claude, Gemini, or whichever).
2. Find the model selector — it's usually a dropdown near where you type your message.
3. Note down every model option available to you. Write down the names, even if they don't mean much yet.
4. Ask the same question to two different models on the same platform: "Explain how a credit card works in three sentences."
5. Compare the responses. Are they identical? Different in tone? Different in detail?

If your platform only offers one model (common on free tiers), that's fine — just note what model you're on and whether upgrading would give you more options.

---

Knowledge Check

1. What is the relationship between ChatGPT and GPT-4o?

• a) They are the same thing
• b) ChatGPT is the platform; GPT-4o is one of the models it uses ✅
• c) GPT-4o is the newer version of ChatGPT
• d) ChatGPT is made by Google; GPT-4o is made by OpenAI

2. Why might a platform offer multiple AI models?

• a) To confuse users into paying more
• b) Because different models have different strengths, speeds, and costs ✅
• c) It's just a marketing gimmick — they all work the same way
• d) Only developers need to know about different models

3. What's the best way to think about the difference between a platform and a model?

• a) A platform is the app or website; a model is the AI engine behind it ✅
• b) A platform is free; a model costs money
• c) They mean the same thing
• d) A platform is for businesses; a model is for personal use