Prompting technique only gets you so far if you're pointed at the wrong model. These 6 lessons cover how model families actually differ, how to choose between them, what changed with reasoning and multimodal models, and the new architecture (world models / JEPA) that doesn't predict tokens at all.
How model families differ
Understand what a 'model family' is and how to pick the right tier for your task instead of defaulting to the biggest model available.
Choosing a model
Work through a practical decision framework for picking a model: input type, task type, budget, then specific capability requirements.
Reasoning models
Understand what a reasoning model does differently, why 'think step by step' isn't necessary advice anymore, and how to prompt reasoning models effectively.
Multimodal models
Learn which current models accept which input types, why that varies by vendor, and how to prompt images, PDFs, video, and audio effectively.
World models and JEPA
Understand the core idea behind JEPA, a model architecture that predicts representations instead of tokens or pixels, and why it matters for physical-world reasoning.
Open vs closed model weights
Understand the difference between open-weight and closed models, the real tradeoffs in cost, privacy, and customization, and when to choose each.
Prerequisites
This track assumes you've used at least one AI model in a chat interface. No architecture or ML background needed — we build up from there.
See current model specs and pricing
