Skip to main content
Back
60 XP

Lesson 10 · 12 min

Capstone: pick the right model for the job

Combine everything: parameters, context, latency, cost, license, and your eval. Decide.

The brief

You're building three different features and have to pick a model for each. Use everything from this path.

  1. A real-time autocomplete in a code editor. Sub-200ms p95. Code-aware. Will run millions of times per day.
  2. An overnight batch summarizer that turns 1000-token meeting transcripts into 2-paragraph summaries. Latency irrelevant; cost matters.
  3. A complex agent that plans 10-step research tasks, calls APIs, and compiles a report. Each task can take 60+ seconds and the answer must be high quality.