Nothing is locked behind a tier.
Every plan has every feature — the validation gate, multi-currency, the git history, the agent. What changes is how much you can do, not what you are allowed to do.
Starter
$0/mo
Free forever. No card, nothing expires.
A real book, not a trial. No card, no expiry.
Start freeSolo
$15/mo
Billed monthly. Cancel whenever.
One person’s books, kept without you thinking about it.
Get Solo →Business
$29/mo
Billed monthly. Cancel whenever.
Small businesses, agencies and bookkeepers.
Get Business →So what does paying actually buy? Today: the work continuing, and support from the person who wrote the thing you are asking about. Not more features — there is nothing held back from the free book, which is unusual enough to say plainly rather than dress up.
That is now changing. The Telegram agent has shipped, and its inference is ours to pay for, so every turn of that conversation is measured. What each plan will include is an open decision rather than a settled one — the prices above are what can actually be charged today, and any change will appear on thebuild log before it reaches a bill.
On every plan, including the free one
- Real double-entry, validated before every write
- Books and collaborators are not counted against your plan
- Multi-currency, investments and cost lots
- Git history — every change a reviewable, revertible commit
- The full ledger UI in your browser
- The HTTP ledger API and the MCP server
- Bring your own agent — the Claude plugin uses your key, not ours
- Nightly off-machine backups, verified by restore
- Your book on its own machine, isolated from everyone else’s
What the same job costs elsewhere.
Read the second row. beancount.io Premium is $14.99 a month, or $8.33 billed annually — the only other commercial hosted Beancount, and on price alone they beat Solo. What you get here instead is a gate that refuses an invalid entry rather than previewing it, a machine per book rather than a shared service, and a roadmap that publishes what we have not built. If price is the whole decision, theirs is the better one, and we would rather say so than hope you do not check. Prices move; QuickBooks and Xero both raised theirs more than once through 2025–26. Treat this as the shape, not a quote.
Including the rows where we lose.
| Countbean | QuickBooks | Spreadsheet | |
|---|---|---|---|
| An agent writes your books | Included, on every plan | Insights, gated to higher tiers | No |
| You own the data | Plain text + git | Locked in their cloud | Yes, but fragile |
| Real double-entry | Yes — refused if it does not balance | Yes | Manual, error-prone |
| Multi-currency and lots | Every plan, including free | Higher tiers | By hand |
| Audit trail | Every change is a diff | Limited log | None |
| Export / no lock-in | Ordinary git repo — but one-click export is not shipped | Painful | Yes |
| Bank feeds | None. You export a file. | Yes | No |
The last two rows are the ones to read twice. One-click export is not shipped — your book is an ordinary git repository and nothing is trapped, but the button that hands you a copy does not exist yet. And there are no bank feeds — which is also why every bank in every country works on your first day, with no integration list to be missing from.