Two ways to pay for models, both in the open. Your keys: never marked up — your provider bills you; our fee is the plan. Prepaid wallet: provider cost + the markup below, reserved before each request and settled to the actual bill. Paid agency plans include a monthly wallet allowance; after that you top up prepaid. No negative balances, no silent fees.
BYOK for up to 2 clients
$0/mo
Run AI for all your clients — beta pricing
$149/mo
White-label everything
$499/mo
Subscription prices are beta pricing — locked in for existing customers when they change. During the beta, new orgs start with conservative managed-wallet limits; contact us to raise them.
Founder-led, fixed-price. For agencies that want the setup done for them.
$299
We configure your org, clients, keys, budgets, and routing policy with you on a call.
$1,500
Full migration from your current provider setup — keys, per-client budgets, reporting, and a parallel-run cutover.
from $3,000
Bespoke routing policies, report templates, and integration work for larger agencies.
Every managed request is priced at provider cost plus exactly this — nothing hidden.
| Plan | Managed markup (direct providers) | Via aggregators (e.g. OpenRouter) | Your own keys |
|---|---|---|---|
| Free | +30% | +37% | +0% |
| Agency | +17% | +24% | +0% |
| Agency Plus | +17% | +24% | +0% |
Why aggregator routes cost more: the aggregator's own fees are inside our cost. Why a markup at all: it's how the wallet, budgets, reports, and support are funded — and it's published so you can always compare against bringing your own key.
Honest math first: if you're one developer buying tokens for yourself, OpenRouter-style direct credits or BYOK through WebikAI at 0% may be cheaper than our managed markup, and we say so. The markup exists for a different job: running AI for other people.
| OpenRouter direct | WebikAI managed (Agency, +17% / +24% via aggregators) | |
|---|---|---|
| Client isolation | one account, shared everything | per-client workspaces, keys, budgets, provider policies |
| Rebilling | export raw usage, build it yourself | white-label monthly report at your own margin |
| Spend control | account-level limit | per-client/project/key caps, reserved before every call; prepaid never goes negative |
| Provider policy | per-request JSON flags | org/client-level: US-only, ZDR-required, allow/deny — enforced, incl. inside OpenRouter routes |
| Quality watch | internal to OpenRouter | published per-provider scores + auto-derank, with receipts on every response |
| Escape hatch | — | BYOK at 0% is always one click away, per client |
Your clients get something too: a monthly report they can read, managed-wallet caps that stop runaway prepaid spend, provider and data-policy rules that are actually enforced, and routing that can avoid providers flagged as degraded. That's what the delta funds — and because the markup is published, you can check the trade any month you like.
Wallet empty → managed requests return 402 with a top-up link;
BYOK keeps working. Minimum top-up $20 (card fees make
smaller amounts a bad deal for you). Balance never goes negative. For streaming responses
where a provider doesn't report final usage, we settle the estimate we reserved and flag the request estimated in your logs — you can see every such request.
Claude/GPT/Gemini are bring-your-own-key by default; wallet access to a frontier model is enabled per endpoint only where our provider terms allow it.
We store request metadata (tokens, model, provider, latency, cost) by default — not prompt or response bodies. See Data handling in the docs.
Yes — block it in your policy; for OpenRouter routes the block is passed through to their inner provider selection.
That's what the quality (beta) scores watch for — and you can pin providers or set a quantization floor regardless.