What's going on, it's Jonathan.
Everybody wants to know which AI to use. I tested them side by side and the answer is annoying: it depends what you're doing. Here's the cheat sheet.
Claude Fable — the worker. Long jobs, real builds, vision. Usage limits are real, so save it for the heavy stuff.
Claude Opus 4.8 — the everyday driver. Where I spend most of my day.
Gemini 3.5 — memory and speed. Big context, fast.
GPT-5.5 — the generalist. Fine at most things, best at none.
Here's what actually matters: the people getting results stack them instead of picking one. Draft in one, build in another, check the work in a third. Picking a favorite is a beginner move.
And there's still one thing none of them can do alone. That's in there too.
Let's build. 🦁
Jonathan
