Key takeaways

  • Run /usage to see session tokens and estimated cost, plus plan usage bars on Pro, Max, Ultra, Team and Enterprise. /cost and /stats are aliases.
  • Run /context to see what is filling the context window; a status line can show it all the time.
  • The dollar figure in /usage is an estimate at list price. Real API spend is on the Console Cost page, and plan spend is under Settings > Usage on claude.ai.

To check token usage in Claude Code, type /usage. It shows the tokens and estimated cost of your current session, and on a paid plan it also shows how much of your plan limits you have used. For what is filling the context window right now, type /context. For a number you can see all the time, set up a status line.

This guide follows the official Claude Code costs page and the commands reference, as of November 9, 2026. I did not run the commands myself, so the sample output below is from the docs.

How to check token usage with /usage

Run /usage at the Claude Code prompt. /cost and /stats are aliases for it, and /stats opens on the Stats tab. The docs describe the command as showing "session cost, plan usage limits, and activity stats."

The Session block at the top shows the current session. This is the sample from the docs:

Total cost:            $0.55
Total duration (API):  6m 20s
Total duration (wall): 6h 33m 10s
Total code changes:    0 lines added, 0 lines removed
Usage by model:
   claude-sonnet-4-6:  1.2k input, 5.3k output, 940.0k cache read, 50.0k cache write ($0.55)

Two things to know about this block:

  • The dollar figure is an estimate. Claude Code computes it locally from token counts at list price. An administrator can set a modelPricing table so it uses your organization's contracted rates, but it is still not an invoice.
  • It covers one session. The totals reset when /clear starts a new session.

The docs also say the Session block is meant for API users. If you are on Pro, Max or Ultra, your usage is included in the subscription, so the cost figure does not matter for billing. Look at the plan bars instead.

After the first response, the block also gets a Prompt cache (main) line. It shows how many requests there were and what share of input tokens came from the cache. It needs Claude Code v2.1.251 or later and covers the main conversation only, not subagents.

How to see your plan usage limits

On a Pro, Max, Ultra, Team or Enterprise plan, the same /usage screen shows plan usage bars and a breakdown of what counts against your limits. The breakdown includes:

  • Attribution: the share of recent usage that went to skills, subagents, plugins and individual MCP servers.
  • Recommendations: suggested changes, each with the share of your usage it applies to, such as trimming a large CLAUDE.md file. Press f to have Claude Code apply the ones it can (v2.1.305 or later).
  • Loops: the heaviest scheduled /loop tasks (v2.1.242 or later).

Press d for the last 24 hours or w for the last 7 days. The docs say the figures are approximate and come from local session history on this machine, so usage from other devices or claude.ai is not included.

If the screen says "Showing last-known usage," the request for your limits failed, most often because the usage endpoint was rate limited. Press r to retry. A message such as "rate limited until 3:14am" means Claude Code cannot fetch your limits until then. It does not mean your work with Claude is limited.

If usage credits are on, /usage also shows a row with your spend this month. Run /usage-credits to manage them. On Pro, Max and Ultra it opens Settings > Usage on claude.ai.

What the plans include

Anthropic's pricing page lists the plans as of November 9, 2026. It gives no numeric limits, only multiples.

Plan Price Claude Code Limits (5-hour and weekly)
Free $0 Not included "Limited usage"
Pro $20 a month, or $17 a month billed annually Included More than Free
Max From $100 a month Included 5x Pro usage
Team Standard seat $25 a month ($20 annual); Premium seat $125 a month ($100 annual) Included Standard: more than Pro. Premium: 5x a Standard seat
Enterprise $20 per seat a month annually, 20-seat minimum, plus usage at API rates Included No figures listed

Anthropic says how fast you hit a limit depends on the model, chat length and file size.

How to see what is using your context

Type /context. It draws the current context usage as a colored grid, with suggestions for heavy tools, memory bloat and capacity warnings. Pass all to expand the per-item breakdown in fullscreen mode. Use it to find which MCP servers or files are taking the space.

Context is not the same as usage. Claude Code sends your full conversation with every request, so a long session keeps drawing on your limits even for a one-line question. The costs page lists the usual causes: long context, cache misses after a break, scheduled tasks and subagents.

How to show token usage all the time

A status line is a script that runs locally and prints text at the bottom of Claude Code. The docs say you can configure it to display context use continuously. We have the full steps in how to set up a Claude Code status line that shows model and token usage.

The status line's cost field shows the same total as /usage. Spend is also visible while Claude works: the line beside the spinner can end with a count such as ↓ 5.5k tokens. That is a live, approximate count of output for the current turn only. Your prompt and history do not move it.

How to check API cost and team spend

Where you look depends on how you sign in.

Your setup Where to see spend
Pro, Max or Ultra /usage and Settings > Usage on claude.ai
Team or Enterprise Spend report in org analytics, with CSV export. It covers usage-credit spend, not usage inside the seat allowance
Claude Console (API) The Console Cost page and the Claude Code dashboard
Amazon Bedrock, Google Cloud or Microsoft Foundry Your cloud billing console

API organizations can set workspace spend limits. Claude Code creates a workspace called "Claude Code" the first time you sign in with a Console account. OpenTelemetry export works on every setup and streams per-user token and cost metrics into your own tools.

For scale, the docs say the average enterprise cost is around $13 per developer per active day, and $150 to $250 per developer per month. The cost is below $30 per active day for 90% of users. For a way to keep a daily cost history, see our piece on Claude Code usage cost, estimates and bills.

When you hit a limit

"You've hit your session limit" and "You've hit your weekly limit" are seat-based windows shared across all models. Switching models with /model does not restore access, and the message shows when the window resets. Run /usage-credits to ask for more, if usage credits are on.

How to use fewer tokens

The docs give these habits for cutting usage:

  1. Run /clear when you switch to unrelated work.
  2. Use /model to pick Sonnet for most coding tasks and keep Opus for complex reasoning.
  3. Lower the effort level with /effort for simple tasks.
  4. Run /mcp and disable servers you are not using.
  5. Keep CLAUDE.md short and move workflow instructions into skills.

We cover more in how to reduce token usage in Claude.

Frequently asked questions

What is the /cost command in Claude Code?

It is an alias for /usage. Both show session cost, plan usage limits and activity stats.

How do I check my Claude Code usage limits?

Run /usage on a Pro, Max, Ultra, Team or Enterprise plan. It shows plan usage bars and a breakdown of what counts against them. Press d for 24 hours or w for 7 days.

Is the cost in /usage what I will be billed?

No. Claude Code computes it locally at list price, so it is an estimate. Pro, Max and Ultra subscribers have usage included in the plan.

How do I see what is using my context window?

Run /context. It shows the current context usage as a colored grid with suggestions for heavy tools and memory.

Sources

  1. Manage costs effectivelyAnthropic
  2. CommandsAnthropic
  3. Customize your status lineAnthropic
  4. Plans and pricingAnthropic

Add ClaudeGPT Si as a preferred source on Google