Credits, not tokens — predictable pricing for AI
Tokens are an internal unit of measurement, not a unit of consumer pricing. We picked credits because they answer the only question that matters: how much will this cost me?
Tokens are an engineering concept. They are how language models internally chop up text. They are useful in a way that bytes are useful — for the people building the system, never for the people using it.
Yet most AI products bill you in tokens. They show you a usage graph in tokens. They tell you a model costs a fraction of a cent per thousand tokens, and then expect you to estimate how many tokens your question will generate. It is like a phone company billing you in megajoules.
MiyoMind bills in credits. One credit is half a US cent of face value. That is the whole pricing model.
The plans
- Free — 100 credits per month, OpenClaw sandbox, Auto mode only, no card.
- Plus — $14.99/month, 6,000 credits, one dedicated container.
- Pro — $39.99/month, 18,000 credits, one dedicated container.
Top-up packs are available on every tier when you want a burst without changing your subscription:
- 600 credits — $3
- 2,000 credits — $10
- 5,000 credits — $25
- 10,000 credits — $50
The ledger never lies
Every credit movement is a row in an append-only ledger. There is no "balance" field that we mutate; the balance is the sum of the rows. This sounds pedantic until you have debugged a billing dispute, at which point it is the only thing you want.
When you send a message, we pre-deduct ten credits and stamp the row with a reservation ID. When the model responds, we compute the actual cost and write a finalising row tied to the same reservation. If the actual cost is less than the pre-deduct, we write a refund row. If a request fails before the model responds, a background reclaim job refunds the full reservation within sixty seconds.
The result: every credit is accounted for. There is no leak between "what you used" and "what you were charged". You can see every transaction, with timestamps, model used, and reason, in your usage dashboard.
Why credits and not dollars?
Two reasons. First, dollars feel weighty in a way that makes people anxious about every message. Credits frame the unit at the right scale — you have 6,000 of them, spend a few without flinching. Second, credits give us a clean abstraction across providers. When a new model arrives or a provider changes their token pricing, we re-tune the credit cost for that model without changing what users see.
A credit is half a cent. Always. The variability lives behind the curtain, where it belongs.
“Predictable pricing is a feature. It is also one of the few features your customers feel even when nothing is wrong.”
If you want to see the full pricing breakdown — including every tool, every plan, and the top-up math — head to the pricing page. If you want to read every credit transaction on your account, the usage dashboard shows the raw ledger.