One of the first usage wallets, and still one of the best.
Claude Code shows how much of your session and weekly limits you've used, when each one resets, and whether you're on track for the week. It doesn't show what a session or a prompt cost against your limit, and the only cap you can set is on extra usage credits.
Overall grade
87/ 100
#1 of 2
In this teardown
At a glance
What it gets right
- Reset times are exact, down to the minute, for both the session and the week.
- /usage estimates where your usage went, broken down by skills, subagents, plugins and MCP servers.
- At the limit you can wait, buy usage credits or upgrade from one menu. Some plans also get a free reset to spend when they choose.
What it misses
- The only forecast is one line saying whether you're on track for the week, and the first warning comes at 85% of the session.
- Limits are percentages of a number Anthropic doesn't publish, and the allowance changed several times in 2026.
- Pro and Max have no usage export, and the breakdown only covers sessions on your own machine.
Enforcement metrics
Max 5x, Oct 1 to Oct 6, 2026. Every test runs three times and we report the median.
Grading detail
01TransparencyWhere did my credits go?85 / 100Counts doubleB
Claude Code shows how much of your five-hour session and your week you've used, with the exact time each one resets, and /usage breaks the session down by model, tokens and tools. What it never shows is what a session or a prompt cost against your limit, so you can't see how your own behaviour moves the bar. The model and effort pickers tell you the relative expense, not the number.
- 1Session and weekly usage, each with an exact reset time.
- 2/usage breaks the session down by model, tokens and tools.
- 3No cost per session or per prompt, so you can't tune your own usage.
What would earn an A
Show what each session and the last few prompts cost against the limit, next to the bar.
02PredictabilityWhen will I run out?84 / 100Counts doubleB
Every limit shows when it resets and a warning arrives before you hit the session cap. The whole forecast is one sentence, "You're on track and should have room to spare", which is better than most tools but no more than that. The effort setting tells you a prompt will cost relatively more, not what it will cost.
- 1Exact reset times for the session and the week.
- 2The on-track sentence is the whole forecast.
- 3Effort shows relative cost, not a number.
What would earn an A
Estimate what a prompt will cost before it runs, and project when the weekly limit will run out at today's pace.
03ControlCan I stop spend I didn't plan?78 / 100Counts doubleC+
The dials up front are strong: you choose the model and the effort, so you decide how fast you burn. Past that, the only limit you can set is a monthly cap on extra usage credits. There is no cap or warning on your included session or weekly usage, so you hit the limit when you hit it. Team plans add per-member limits, which we didn't grade here.
- 1Model and effort pickers control how fast you burn.
- 2A monthly cap on usage credits is the only limit you can set.
- 3No cap or alert on included session or weekly usage.
What would earn an A
Let me set a ceiling on included usage, warn me when I'm tracking past it, and give each agent its own budget.
04Top-upsHow fast can I get unblocked?94 / 100A
Buying more usage is simple: pick $100, $250, $1,000 or a custom amount in Settings › Usage, and it lands in your wallet with a small discount at the larger sizes. At a limit, one menu offers wait, buy usage credits or upgrade, and eligible plans get a free reset. The one gap is that the purchase is in dollars, so you can't tell how much usage you just bought.
- 1Buy $100, $250, $1,000 or a custom amount in a couple of clicks.
- 2Wait, buy credits, upgrade or use a free reset from one menu.
- 3Purchases are in dollars, with no sense of how much usage that is.
What would earn an A
Show what a purchase buys in usage, and offer auto-reload right on the limit screen.
05Real timeDoes the meter move when I do?95 / 100A
The usage bar moves as you work and the limit alert appears in under a second, enforced on the request itself rather than by a webhook or an email after the fact. The context-window indicator updates the same way. Credits bought on the web showed up in the terminal after a refresh.
- 1The bar and the limit alert update in under a second.
- 2Limits are enforced on the next request.
- 3Credits bought on the web appeared after a refresh.
What would earn an A
Update the balance everywhere within a second, including the billing page.
06FairnessDo the rules feel fair?98 / 100A+
Extra usage costs the same as plan usage, and the free reset is the most generous touch we've seen in a wallet: if you're burning through the week too fast, one click refills the session and the week. Everything about it is built to stay out of your way. Limits are a share of a number Anthropic doesn't publish, which is the one thing that keeps it from perfect.
- 1A free reset refills the session and the week when you choose.
- 2Extra usage costs the same as plan usage.
- 3Limits are a share of a number Anthropic doesn't publish.
What would earn an A
Publish what the limit is, and give a month's notice before it changes.
07DesignIs it clear and close to the work?93 / 100A
Usage is one command away in the terminal and one click away in the apps. The panel is two bars and two reset times, and the limit message tells you exactly when you can continue.
- 1/usage is one command from the prompt.
- 2Two bars, two reset times.
- 3The limit message says when you can continue.
What would earn an A
Raise the contrast on the bars, and make every control reachable from the keyboard.
How the Claude Code wallet works
Checked Oct 14, 2026
- Unit of usage
- Percent of a five-hour session limit and weekly limits
- Plans tested
- Pro ($20/mo) and Max 5x (from $100/mo)
- Limit windows
- Rolling five-hour sessions, plus weekly caps for all models and per model
- Where you see it
- /usage in the terminal, Settings › Usage on web and desktop
- At the limit
- Wait for the reset, buy usage credits, upgrade, or redeem a free reset
- Usage credits
- Prepaid at API rates, with a monthly limit and optional auto-reload
- Upgrade path
- Pro to Max, at 5x or 20x Pro's usage
- Team controls
- Spend limits by org, seat tier and member; members can request more
- Usage export
- None on Pro and Max. Daily CSV on Team, Analytics API on Enterprise
- Rollover
- None for plan limits. Usage credits generally don't expire.
Three best-in-class elements
What Claude Code does best, and how to build each one with Schematic.
Resets in 2 hr 14 min
The reset clock
Put the exact time capacity comes back next to every limit, in the user's own timezone.
Seen in Claude Code, Figma
With Schematic: Limits that reset daily, weekly or monthly, shown in your app with usage components.
A free reset you can save
Eligible plans get a one-time reset that refills the session or the week, spent whenever the customer chooses.
Seen in Claude Code
With Schematic: A one-time credit grant with its own expiry, issued to a single company.
A paid escape hatch with a ceiling
At the limit, people keep working on usage credits, capped by a monthly amount they chose.
Seen in Claude Code, Cursor
With Schematic: Overage or auto top-up past the plan limit, with a hard limit that stops spend where you set it.
How other AI coding wallets compare
See the full rankingHow we tested
We paid for Pro and Max 5x and used both for six working days of real feature work. Every grade comes from the public rubric, version 1.
Disclosure
Schematic builds wallet infrastructure. We disclose any commercial relationship with a company we review.
Build a wallet your customers expect.
Schematic puts credits, metering and limits on one real-time ledger, with wallet components your customers see inside your product.