Pay per token. Watch it happen.

Apertureis prepaid: you top up a credit balance, and every request charges exactly what it used — input tokens at the model's input price, output tokens at its output price. No subscription, no minimum, no expiry games. The meter runs while the answer streams.

Live meter demo

Watch the meter run — no account needed

gpt-oss 20B · capped at 200 tokens · gpt-oss-20b

How billing works

Micro-dollar precision

Charges are computed in integer micro-USD (1 USD = 1,000,000 µUSD) and rounded once, at the end. A 400-token request on a cheap model costs a fraction of a cent — and you're charged that fraction, not a rounded-up cent, not zero.

$0.20 free on signup

Every new account gets a one-time $0.20 grant — about 1,000,000 output tokens on gpt-oss 20Bat today's prices. Enough for a real conversation before you pay anything.

Top up $5 at a time

Add credits in $5 packs (≈ 25,000,000 output tokens on gpt-oss 20B). Your balance is a ledger, not a subscription: it only moves when you use it, and every movement is itemized on your usage page.

What requests actually cost

Computed live from gpt-oss 20B's current prices — $0.05/M in · $0.20/M out — with the same rounding production billing uses.

Example request costs on gpt-oss 20B
RequestInput tokensOutput tokensYou pay
Quick question150300$0.000068
Long coding turn2,0001,200$0.00034
A 30-message session20,0009,000$0.0028

Prices vary by model — see the full catalog. Aborted requests are charged only for the tokens actually generated.

The meter starts at $0.00

Get started with $0.20 free

prepaid · per token · every charge itemized