Sats/token

01 / the problem

A token is irreversible. A card payment is not.

Generating a token draws joules from the grid and turns them into heat and part of a response. No message from a card network can reach back and recover that energy. So if you sell inference on cards, you are a lender, not a merchant.

You allocate scarce GPU time now, for an amount you do not know in advance, and learn somewhere between 60 and 120 days later whether you were paid. Each familiar barrier is an attempt to underwrite that loan, and the fraud you fail to catch is the default rate.

1. Sign-up wallAn account before the first token.
2. Fraud checksInvasive verification for a few cents of usage.
3. Minimum top-upYou wanted to try a model, not prepay a month.
4. Unsupported countriesThe rail decides who may buy.
5. Keys tied to a humanSomeone who can be charged later.

Fallible defenses stand between a payment that can be undone and a product that cannot.

Pick a rail whose properties match the product.

02 / how it works

The payment is the credential.

The server answers an unpaid request with HTTP 402: a macaroon that commits to an invoice's payment hash, plus the invoice. Paying reveals the preimage, and macaroon plus preimage is the credential. A valid credential cannot exist without settlement. No sign-up, no key issuance, no account.

pi-402 takes a Payment offer when the seller makes one and falls back to L402. It accepts the charge intent by default; session needs a return-invoice minter that the swap-backed wallet cannot supply yet.

03 / metering

Charge for the thing that varies.

One request costs the seller a few dozen tokens, the next a few thousand, and the seller only learns which once the response is served. So the payment happens once, up front, for a bundle, and the accounting happens per request, from the usage the upstream actually reported. At 1 msat per input token and 2 msat per output token, that is roughly $0.79 / $1.57 per million tokens at the time of writing.

The four calls below are the same chips that light up in the meter at the top of the page. Run a completion there and watch them fire.

Aperture ships meterd. A real seller wants lightmeter.

meterd is the minimal price server: a JSON file of per-model rates. Lightmeter is the reference daemon for a real seller. It knows what each model costs, reads the upstream provider's catalog, tracks the BTC/USD rate, converts dollars per million tokens to msat per token, adds an optional margin, and keeps a SQLite database with an append-only accounting ledger for reconciliation. Try the conversion with the values from the regtest demo config.

04 / wavelength

No node. No channels. Nothing to keep online.

Running a Lightning node means channels, inbound liquidity, and a process that has to stay online. Too much to ask of a seller who wants to sell tokens, and far more than you would ask of a wallet living inside a coding agent. Wavelength is an Ark wallet that pays a BOLT11 invoice through a swap server, so both sides skip it.

05 / agents

Agents pay for their own inference.

Plenty of agents can pay a Lightning invoice for an API. Here the metered thing is the model call itself, priced by the token and settled by the preimage. pi-402 is one example: an extension for the Pi coding agent that registers a paid model provider. Pi is unmodified.

06 / run it yourself

Bring your own seller.

There is no hosted endpoint at launch. You run aperture and a price server in front of any OpenAI-compatible upstream, then point a buyer at it. Config below is trimmed from the regtest demo profile; the READMEs have the full steps.

# challenge invoices come from a Wavelength wallet, not lnd
listenaddr: "127.0.0.1:8404"
authenticator:
  network: "regtest"
  wavelengthgateway: "https://your-wavelength-gateway.example"   # your gateway; the regtest demo runs one locally
services:
  - name: "inference"
    hostregexp: ".*"
    pathregexp: "^/v1/.*"
    address: "inference.baseten.co"   # any OpenAI-compatible upstream
    protocol: "https"
    auth: "on"
    headers:
      Authorization: "Bearer ${BASETEN_GLM_5_2}"
    dynamicprice:
      enabled: true
      grpcaddress: "127.0.0.1:10025"
      insecure: true
      metered: true

listenaddr: "127.0.0.1:10025"
dbpath: "lightmeter.regtest.db"
defaultmodel: "moonshotai/Kimi-K3"
bundletokens: 1000000
models:
  "moonshotai/Kimi-K3":
    inputusdpermtok: 3.00
    outputusdpermtok: 15.00
pricing:
  marginpercent: 25
  btcusd:
    staticrate: 100000   # pinned for regtest; 0 polls spot

go build ./cmd/lightmeterd
./lightmeterd --config lightmeter.yaml

pi install npm:@lightninglabs/pi-402
export L402_PROXY_URL=https://your-seller.example/v1   # your seller; include the /v1 prefix

/fund 20000                       # invoice; pay it from the Wavelength app
/model l402-baseten/kimi-k3
# then prompt as usual. /balance and /receipts show the ledger

export PI402_MAX_SAT_PER_REQUEST=...   # per-request ceiling
export PI402_DAILY_CAP_SAT=...         # daily ceiling