I’ve read that some platforms give you access to 400+ AI models through a single subscription. That sounds incredible in theory, but it also sounds overwhelming in practice. How do you actually choose which model to use for each step of your browser automation?
Like, for extracting product prices from a page, do you use GPT-4, Claude, Gemini? Do you test each one and compare results? Do you use different models for different steps, or stick with one? And how much does model choice actually affect automation quality versus cost?
Isn’t having too many options paralyzing? Or have people found a framework for making these decisions?