32dots HEIDELBERG AI
Session 6 easy

Models, usage & cost

LESSONLesson 6 · ~15 min

🎯Goal. Pick the right model for the job and keep an eye on what you're spending.

  1. 1Use /model to switch mid-session. Sonnet handles most coding well and costs less; reserve Opus for complex architectural or multi-step reasoning. Set a default in /config.
  2. 2Check spend with /usage — on a paid plan it breaks usage down by skills, subagents, and MCP servers, with a 24-hour / 7-day toggle.
  3. 3Plans: Pro ($17/mo annual, $20/mo monthly) includes Claude Code; Max (from $100/mo) gives 5× or 20× more usage. On API/Console accounts you pay per token — enterprise deployments average around $13 per developer per active day.

You'll see. A clear read on which model fits a task and where your usage is going — Sonnet for day-to-day, Opus when the problem is genuinely hard.

💳Cost. Model choice is the biggest lever — Sonnet for most work, Opus only when needed; /usage keeps spend visible.

💡Takeaway. Default to Sonnet, switch to Opus for hard reasoning, and watch /usage so cost never surprises you.