Documentation
Everything you need to plug CheapAI into your tooling — setup guides for popular agents, API reference, FAQ.
-
Compact your history
Long context is expensive. Run
/compactin Claude Code to compress the history — especially before switching tasks. -
One key per project
Create a separate key for each project — easier spend tracking and a quick revoke if one leaks.
-
Watch your spend
The Logs page shows every request — model, tokens, cost, duration.
-
Pick the right model
Haiku / Flash — light mechanical work and bots. Sonnet — workhorse for agents. Opus — heavy reasoning when quality matters.
-
Use the cache
System prompts and stable documentation are cached automatically by Anthropic. A long session with one prompt saves 70-90%.
-
/cost ≠ actual spend
Claude Code's built-in
/costuses Anthropic's official pricing — we're 30-40× cheaper. Look at Logs for the truth.