Use /cost to Track Token Usage and Spending During Your Session
Long sessions with big codebases can burn through tokens quickly. Instead of guessing how much context you're using or what your session is costing, just check.
/cost
Claude Code shows you the token breakdown for your current session — input tokens, output tokens, total cost, and how much of the context window you've used. No surprises on your bill.
This is useful at several points during a session:
# Before a big task — check your starting point
/cost
# After Claude reads many files — see the context impact
/cost
# Before deciding whether to /compact or /clear
/cost
# At the end of a session — see what it cost
/cost
The information helps you make practical decisions:
- Context is filling up? — run
/compactto free space, or/clearif you're switching tasks - Costs higher than expected? — switch to a cheaper model with
/modelfor the remaining work - Big file reads eating context? — be more specific about which files Claude needs to read
- Session getting expensive? — break the remaining work into a new session
A few patterns for cost-conscious usage:
# Start with haiku for exploration, switch to opus for the hard part
claude --model haiku
# explore the codebase, understand the problem
/cost
/model opus
# do the complex work
/cost
# Check cost before and after a big operation
/cost
"Read all the models and generate an ER diagram"
/cost
For teams tracking AI spending, /cost gives you the data you need to understand where tokens go — large file reads, complex multi-step tasks, or long conversational sessions.
You can't optimize what you don't measure — /cost shows you exactly where your tokens go so you can spend them wisely.
via Claude Code
Log in to leave a comment.
When Claude starts referencing files you've since deleted, remembering old code you've already changed, or getting confused by contradictory instructions from a long session — type /clear to wipe the slate clean. Unlike /compact which preserves context, /clear gives you a true fresh start without restarting the CLI.
Not every task needs deep reasoning. Type /fast to switch Claude Code into fast mode — same model, faster output — for quick edits, simple questions, and routine changes. Toggle it off when you need Claude to think harder on complex problems.
Long sessions eat through your context window as conversation history piles up. Type /compact to summarize the conversation so far and reclaim space — keeping Claude's understanding of what you're working on while freeing up room for more work.