
Claude Usage Credits Expire: The Six-Month Rule Explained
Claude usage credits now expire six months after purchase in certain jurisdictions such as Japan, starting September 10, 2026. What the policy says, what it means for prepaid habits, and how to budget credits vs. subscriptions.
Prepaid credits have one job: to sit there until you need them. That assumption just changed for at least some Claude users. According to the Claude Help Center article on managing usage credits for paid plans (article 12429409, last updated around August 10, 2026), "in certain jurisdictions such as Japan, usage credits expire six months after purchase starting September 10, 2026," with affected users receiving an email notice.
That's one sentence with a lot of load-bearing ambiguity in it. Let's take it apart carefully.
What the policy actually says
Sticking strictly to what's verifiable:
- The expiry clock is six months from purchase — not six months from when the credit is applied to your account, and not from first use.
- The start date is September 10, 2026. The help page was last updated around August 10, 2026, and affected users get an email.
- "Certain jurisdictions such as Japan" is the exact scope language. The article names Japan as an example of the jurisdictions affected; it does not, in the language we could verify, publish a complete list. So the honest statement is: if you're in Japan, the six-month expiry applies to you from September 10, 2026; if you're elsewhere, check your email and the help center rather than assuming either way. Hedge on jurisdiction is deliberate here — no source we found enumerates the full set of affected regions.
One plausible reading — again, a reading, not a confirmed fact — is that this reflects local consumer-prepayment or gift-credit rules in specific markets rather than a global Anthropic policy change. Japanese prepaid payment regulation is unusually specific about unused balances, which would explain why Japan is the named example. But Anthropic's page doesn't say that, so don't repeat it as though it does.
Connect the Claude or Codex you already pay for — the rest runs on workers that cost a fraction.
Download meshcode →What it means for prepaid habits
Six-month expiry quietly changes the economics of buying credits in bulk:
- Bulk discounts stop being free options. If you bought a large credit pack intending to use it "someday," a hard expiry converts unbuilt months into a total loss. Credits only beat a subscription if you actually consume them inside the window.
- The purchase trigger changes. The rational move under an expiry rule is to buy credits shortly before spending them, not during a sale. That's the opposite of how most people treat prepaid balances.
- The expiry is invisible until it isn't. Nothing in your daily workflow surfaces "this balance dies in 19 days." If you're in an affected jurisdiction, a calendar reminder six months minus a week after each purchase is unglamorous and sufficient.
None of this makes prepaid credits a bad idea. It makes stale prepaid credits a bad idea. That distinction is the whole post — and it's the same distinction we work through in prepaid credits vs. subscription for AI coding.
Credits vs. subscription: how to budget now
The two shapes fail in opposite directions, and an expiry policy sharpens the difference:
| Prepaid credits | Monthly subscription | |
|---|---|---|
| If you use a lot | You pay for what you burn — no waste | Flat price, effectively discounted |
| If you use a little | Balance can expire unused (new risk) | You pay full price for a quiet month |
| Predictability | Hard ceiling — you can only spend what you've added | Recurs regardless of use |
| Expiry risk | Six months in affected jurisdictions | None — access stops when you stop paying |
Practical rules for the next few months:
- Subscribers: usage credits added on top of a plan are the ones with the new clock. Before topping up, check whether your current plan's allowance actually covers the gap — making a Claude Pro subscription last longer is mostly about pacing and model routing, both of which are free.
- Credit buyers: buy in smaller, more frequent increments until the full jurisdiction list is public. If the policy does turn out to cover your region, you've lost nothing; if it doesn't, you've lost nothing either.
- Comparison shoppers: credit expiry is one more line item in the Claude vs. Codex cost comparison — a 5%-cheaper credit pack with a 0% survival rate past six months is not cheaper than anything.
See the pricing yourself: meshcode is free to start — download it
The honest bottom line
The verifiable facts are narrow: six months, from purchase, starting September 10, 2026, in certain jurisdictions such as Japan, with email notice. Everything beyond that — which other regions, why the rule exists — isn't published yet, so plan around the jurisdiction you can confirm rather than the one you hope you're in. And when you do load a balance anywhere, the healthiest habit is the oldest one in the book: only prepay what you're confident you'll spend before the clock runs out. meshcode's model takes that seriously by design — free to start, top-ups from $1, prepaid so you can only spend what you've added, with a 5% + $0.50 flat fee at top-up and tokens billed at cost.
👉 Download meshcode — Mac, Windows
More from the blog
How to Run Claude Code and Codex in Parallel on One Repo (Git Worktrees, Step by Step)
Run Claude Code and Codex on the same repository at once without them overwriting each other: one git worktree per agent, scoped briefs, and a safe merge. Step by step.
The Cheapest Way to Use Grok for Coding in 2026
The cheapest way to use Grok for coding depends on how you work: API tokens from about $1 per million, Cursor Pro at $20 with a dedicated Grok pool, or SuperGrok at about $30. What each route really costs.
Can You Use Claude Code and Codex at the Same Time? (2026 Guide)
Yes. Claude Code and Codex have separate logins, billing and usage limits, so you can run both on one project today. How to set it up, what goes wrong, and what it costs.