Model Landscape for AI Engineers
Model landscape is the working knowledge of which model families exist and how to choose among them: frontier hosted models, open-weight models you can run yourself, and small models for cheap or on-device use. It sits early on the AI Engineer path so that model choice rests on model cards and your own tests, not marketing.
Stage: Foundations · Part of the AI Engineer roadmap · Last reviewed 19 November 2026
What to learn in Model Landscape
- Compare hosted, open-weight and small models on cost, latency, privacy and quality
- Read a model card for context length, licence, training data notes and known limits
- Treat public benchmark scores sceptically, noting contamination and mismatch with your task
- Build a small task-specific test set to compare two candidate models fairly
- Check a licence before deciding whether open weights can be used commercially
Model Landscape interview questions
Why can a model that tops a public leaderboard still perform worse in your product?
Benchmarks measure fixed tasks that may not match your domain, prompts or data format, and test questions can leak into training data. Scores also hide cost and latency. A small evaluation set built from your own real inputs is a better basis for choosing.
What does open-weight mean, and how is it different from open source?
Open-weight means the trained parameters can be downloaded, so you can self-host and fine-tune, but the training data and code may stay private and the licence may restrict use. Open source in the strict sense grants broad rights to use, modify and redistribute. Read the specific licence.
Where Model Landscape fits in the AI Engineer roadmap
Model Landscape belongs to the Foundations stage (beginner level), which takes about 72 hours. What you need before touching models.
Other units in this stage
Back to the full AI Engineer roadmap