← Read the full guide
All infographics

Claude at Maximum Efficiency

5 models. 5 thinking modes. 30+ slash commands. The toolkit most Claude users barely scratch.


5
Models
5
Thinking Modes
30+
Slash Commands
4
Plan Tiers
Pick the right model
FLAGSHIP
Opus 4.7
Hard reasoning, architecture, code review. Most expensive — use when quality matters.
Opus 4.7 1M
Same Opus, 1M token context. Long sessions + large codebases only. Pricey.
Long Ctx
Sonnet 4.6
Default for most coding. 80% of Opus, 5× cheaper. Workhorse.
Default
Haiku 4.5
Fast, cheap, surprisingly capable. Batch + high-volume + automation loops.
Speed
Thinking effort & daily commands
Thinking Modes
  • Low — greetings, quick edits
  • Medium — chat, drafts, simple code
  • High ★ default for serious dev
  • Extra high — hard debug, architecture
  • Max — rare; expensive; "I'll wait"
5 Slash Commands You'll Use Weekly
  • /usage — see your quota
  • /context — how full is the window
  • /compact — summarize & continue
  • /clear — fresh start, same project
  • /model — swap mid-session
Pro Tip
Default to Sonnet. Promote to Opus only when the task is hard enough to justify 5× the cost.
Common Mistake
Adding 50 MCP servers. Each one costs context window before the conversation starts. Keep 5-8 daily-use; install niche ones per-project.
Daily playbook
When to upgrade plan
  1. Open Claude Code in project root
    — CLAUDE.md auto-loads
  2. Pick model intentionally
    Sonnet default, Opus for hard work
  3. State the goal up-front
    Claude works better with a stated goal
  4. Watch /context at 60%
    Compact before you hit the wall
  5. End: update CLAUDE.md
    Decisions + conventions for next session
  1. Free → Pro $20
    When Free cap blocks your day
  2. Pro → Max $100
    When Pro caps hit weekly
  3. Max $100 → Max $200
    When /usage hits 80% daily