Module 2: The Main Roads — The Creative Companion

Most people think they're choosing a tool. They're actually choosing a model type inside a tool.

That sounds more complicated than it is. Here's the cheat code: every major AI platform runs multiple models under the hood, and those models fall into a handful of categories. Once you see the pattern, the whole landscape snaps into focus.

double-click any line that resonates

// the cheat code

Four Types, Everywhere

Instant models are built for speed. Fast drafts, quick answers, rapid back-and-forth. You reach for these when you're brainstorming or just need something now.

Thinking models slow down on purpose. They reason through multi-step problems, weigh trade-offs, and produce more considered output. When the question is hard, this is the gear you want.

Multimodal models handle more than text — images, voice, sometimes video alongside the words. The input gets richer, so the output can too.

And Integrated models live inside the apps you already use. Your documents, your email, your spreadsheets. They don't ask you to go somewhere new. They show up where you already work.

Same four types. Every platform. Once you see it, you can't unsee it.

// the big four

The Players

ChatGPT

OpenAI — San Francisco, founded 2015

The one that started it all. GPT-3 landed in 2020 and was the first time most people saw AI that could actually write. Then ChatGPT went public in late 2022, millions signed up in the first week, and suddenly AI wasn't theoretical anymore — it was something you could talk to.

The current flagship is GPT-5.2, and it comes in flavors. The Instant variant handles fast writing and everyday prompts. The Thinking variant goes deeper on harder problems. Pro tackles heavier professional workloads. And Auto — the smart default for beginners — picks the right variant behind the scenes so you don't have to think about it.

As of February 2026, OpenAI retired GPT-4o and the older models entirely. That's a big signal: 5.2 is the main lane now.

best for

Building outlines and systems. Turning messy ideas into clear steps. Planning, structuring, synthesizing.

the feel

Like a sharp analyst who always has a framework ready.

Claude

Anthropic — San Francisco, founded 2021

Built by former OpenAI researchers who wanted to take a different path — safety and alignment first, capability built on top. Claude 1 launched in 2023 focused on being helpful and honest. The models kept evolving through Claude 2 and the Claude 3 family, each generation getting sharper without losing the warmth.

Anthropic uses a family naming system that stays consistent even as the technology advances. Haiku is the instant model — fast and lightweight. Sonnet is the balanced workhorse for everyday creative work. Opus is the deep thinker for complex, heavy-lift projects. The latest, Opus 4.6, shipped in February 2026 with improved long-task reliability and massive context windows — meaning it can hold an entire book-length project in its head at once.

best for

Writing with tone control. Editing long drafts without flattening your voice. Thoughtful critique and refinement. Long-document work where the AI needs to hold the whole picture.

the feel

Like a thoughtful creative partner who listens before responding.

// philosophy

How They're Different

OpenAI and Anthropic share roots but different instincts. OpenAI pushed for capability first — how powerful can we make it? Anthropic focused on safety and alignment — how do we make it trustworthy?

For creative writers, the difference shows up in how they respond. ChatGPT tends to give you the answer. Claude tends to give you the conversation. One hands you the fish. The other sits down and asks what you're hungry for.

Neither is wrong. They're different instruments.

Gemini

Google / DeepMind

Google's AI lives in two places: the standalone Gemini app and deep inside the ecosystem you probably already know — Search, Docs, Gmail, Workspace. The model lineup splits into Flash for speed and Pro for depth, with Gemini 3 Flash currently serving as the default in the consumer app.

The real strength is integration. If you already live in Google's world, Gemini meets you there without asking you to change a thing.

best for

Research and well-sourced answers. People already living in Docs, Gmail, and Workspace. Multimodal work where you're combining images, audio, and text.

the feel

Encyclopedia-smart. Useful. Less conversational.

Copilot

Microsoft — Powered by OpenAI's models

AI built directly into the Microsoft products you already know — Word, Excel, Outlook, Teams, Windows itself. Not a separate destination. An assistant embedded in your existing workflow. Microsoft recently brought GPT-5.2 into the Copilot stack, and the current push is toward Agent Mode — moving beyond "help me write" into "help me do."

best for

Drafting and rewriting inside documents. Summaries from emails, meetings, and files. Excel wizardry with tables, formulas, and insights. Company workflows where permissions and governance matter.

the feel

A capable assistant who already knows where everything is.

// the pick

If You Freeze at the Menu

Clean structure and planning? ChatGPT. Writing, voice, and long drafts? Claude. Google integration and quick research? Gemini. Office work done inside Office? Copilot.

No loyalty required. These are instruments, not identities. Many people use more than one.

For this course — the kind of creative work that lives or dies on voice, nuance, and honest feedback — we'll focus on Claude.

But knowing the full landscape means you're never stuck with just one option. That's the point.

logged

where you are

NowYou know the Big Four — and the model types underneath them all.

NextA quick look at the specialists — smaller tools worth knowing about.

ThenYou'll set up your account and type your first prompt.

Continue to Module 3

The Side Roads

Specialists outside the Big Four.

→