I'm a software engineer with 5+ years of industry experience. Together, we're going to go through a primer on all these AI terms and what you should know about them — starting at a high level and then going deeper.

There are two dominant new players in this space right now: OpenAI and Anthropic. You can think of them like Google and Apple — not necessarily in terms of their values or ethics, but simply as two very big and prominent companies in tech right now.

OpenAI has a chatbot called ChatGPT, and Anthropic has a chatbot called Claude. This is similar to how Google has Chrome and Apple has Safari — they're just the names for each company's chatbot offering.

Note: Google has its own chatbot offering called Gemini, which is another big player in this space.

Each of these chatbots is powered by an LLM, which stands for Large Language Model. All of the large language models developed by OpenAI have "GPT" in their name — GPT-4o, GPT-4o mini, GPT-5, and so on. On the Anthropic side, the models have different names: Opus, Sonnet, Haiku. The naming is a little more involved than the GPT style, but it's the same concept.

This is very similar to software updates on a phone. If you had an iPhone, you might have been on iOS 17 back in the day but are now on something much newer. Each model version increment — and on the Anthropic side, each new model name — indicates new capabilities and more power.

You might remember when ChatGPT first came on the scene, everyone would say, "It's really cool, but it hallucinates a lot." Now you might notice the hallucination rate is significantly lower. The reason for that is that they've upgraded the model powering the chatbot.

When you look at ChatGPT and Claude, there's basically a website interface wrapping the large language model that's being hosted under the hood.

The reason new model releases make a lot of news is because developers all over the world — working on startups or at big companies — are paying close attention. Each model release brings new capabilities, which means new things you can build and add to your app that you couldn't do before. And because this is state-of-the-art technology, some of these things have never been possible at any point in history.

The parallel I'll draw: before GPS navigation came to mobile phones, there was simply no way to do what Uber and Lyft do now. When that capability was released, it was major news, because it unlocked entirely new categories of applications. We're all used to having GPS in our pockets now, but at the time it was a genuine step change. New AI models work the same way — they unlock things that simply weren't possible with earlier versions.

Say you have a financial app that takes people's financial data, performs analysis on it, and gives recommendations on budgets. If you built this with the original GPT model or the original Claude model, your ability to perform reliable analyses would have been limited — there would be hallucinations and inconsistent quality. Now, the quality you can surface to users is dramatically better. That's why, as a developer, you actively follow model releases: it directly impacts your ability to deliver value to your customers.

This topic is one that I get a lot of questions on from friends and family, so I wanted to put it out there. If this is new to you, that's totally fine — this space is changing really fast and there's a lot of complicated terminology. There's a joke in software engineering that the hardest thing to do is name things, and I think we see that a little in how Anthropic names their models 🙂