Here’s where it gets ridiculous: o3-mini launched in January, three months before o3 itself. That’s like releasing the iPhone mini before announcing the iPhone exists.

I’ve been trying to keep track of their lineup: GPT-4, GPT-4o, o1-preview, o3-mini, o3, o3-pro. What does any of this mean? The “o” supposedly stands for reasoning optimization, but good luck explaining to a client why they need o3-pro instead of GPT-4o for their use case. Developers are stuck choosing between gpt-4o, o1-preview, and now three flavors of o3 in their API calls. Each has different pricing and capabilities, but the names tell you absolutely nothing about what you’re getting. Meanwhile, Apple just goes iPhone 14, iPhone 15, iPhone 16. Revolutionary concept: numbers that go up.

The worst part? This actually matters when you’re building applications. Is o3-mini better than o1? How does o3 compare to GPT-4o? Nobody knows without diving into technical documentation that probably doesn’t exist yet.

OpenAI went from the beautifully simple GPT-3, GPT-4 progression to whatever fever dream this has become. At this point, I’m expecting them to release GPT-π next, followed by GPT-√2-ultra-mini-pro.

Update (August 2025): Plot twist—GPT-5 is actually called GPT-5. Sometimes the universe makes sense after all.