Ainfera
03 · Pricing

A share of the savings.
Never a markup.

Your all-in is always below calling direct. If a month isn't, the difference is free.

Provider cost passes through. You pay a share of the savings routing proves — never a markup, never a per-call fee. No seats, no tiers, no sales call.

Free during beta · payment turns on at GA.
You're billed only for inference. The savings-share rate is held until billing computes at GA — the page quotes no rate yet.
spend calculator

Plug in your monthly numbers.

Three inputs. See what your agent costs going direct versus its Ainfera-routed cost — the savings-share rate is held until GA, so the panel quotes no rate.

monthly calls100,000
1k 2M
avg tokens per call1,700
200 8k
if you went directclaude-opus-4-7
we'd route this workload to a mix · see right
If you called claude-opus-4-7 directly$3,621.00
Ainfera-routed direct cost$314.50
+ Ainfera · share of proven savingscomputes at GA
You payrouted cost + a share of the savings
Your all-inis always below calling direct. If a month isn't, the difference is free.
vs going direct on claude-opus-4-7. The exact savings-share rate computes at GA; this panel quotes no rate. Provider list prices are illustrative — your real workload routes across a mix of models per call.
funding & payment

Fund in USDC. Or just use a card.

USDC is the native rail — settled per call over x402 on Base, the same chain every routing decision is logged to. Prefer fiat? A card tops up the same workspace balance through Stripe, which also takes USDC via Pay-with-Crypto. One balance, settled the same way — you don't need crypto to use Ainfera.

Payment turns on at GA.
Beta is free. When billing computes, USDC settles first; card and USDC top-ups through Stripe follow under our US entity. We'll give 30 days' notice before anything is charged.
questions

A few we get a lot.

When does payment turn on?
At GA. We'll tell you 30 days ahead, in the dashboard and by email. Beta is free — your beta usage isn't billed retroactively.
How is the charge computed for a fallback call?
One settled call, one charge — a share of the savings, regardless of how many candidates we tried. Failures don't count. You only pay on a settled call.
Do I need crypto, or can I pay by card?
Either. USDC is the native rail (x402 on Base), settled per call. Cards — and USDC top-ups — run through Stripe into the same workspace balance, so a card alone is enough. Payment turns on at GA; beta is free.
Can I cap monthly spend?
Yes — caps are per agent and per workspace. Hit the cap and we pause routing, log the event on chain, and post you a webhook. Nothing surprise-bills.
next

See the models we route across. Models →

routing · activeblock #12,273models · 245audit · on-chainainfera · the inference of ai agents