Pricing
Prepaid balance. Money held before the call, settled after it.
There is no plan to choose and no seat to buy. You top up a balance, your agent spends against it one call at a time, and every cent that moves is an entry in an append-only ledger you can read.
What you are charged, in one sentence
Omnial MCP holds the top of the estimate before the call and then charges what the call actually cost. Below the quote, the difference is released back to available. That is the ordinary case, and the one nothing about this change touches. Above it, the excess is charged too, which can take your balance negative; that is a debt, and no new run starts while it stands. A hard ceiling of 2x the quote is enforced on every single run: past that we absorb the difference and pull the tool out of the catalog on the spot, because a call that costs a multiple of its own quote is our defect rather than your usage. A run that fails is not charged at all.
The model
What you are charged, and when.
You top up a balance
A card payment through Stripe credits the workspace balance. The minimum top-up is $5.00. Optional auto top-up refills the balance when it falls below a threshold you set, with the refill required to be larger than the threshold, a cooldown after a successful charge, and a hard daily ceiling on automatic charges, because a refill that lands you back under your own threshold is a card in a loop.
The quote is held while the call runs
Before the provider is contacted, the run writes a hold for the top of its estimate and moves that amount from available to held. If that hold would exceed your available balance or a spend cap, the run is refused before anything is called and nothing is charged.
A held amount is a quote, not a cap on the charge. It is what a call is expected to cost, reserved so the money is there when the bill arrives.
Settlement charges what the call actually cost
When the run finishes, the real cost is computed from the most authoritative source that tool has, never blindly from the estimate, and charged, in one transaction with the release of whatever is left of the hold.
Under the quote, that means what it always meant: you pay the smaller number and the rest comes straight back. Over the quote, the excess is charged as well.
A charge above the quote is bounded, and visible
A call can cost more than it was quoted: an estimator that was wrong about a quantity, a provider that billed for more than it was asked for. That difference is yours, and it can take your balance negative; a negative balance is a debt and no new run starts while it stands.
It is bounded on every single run at 2x the quote. Past that we absorb the difference and pull the tool out of the catalog immediately, because a call that costs a multiple of its own quote is our defect and not your usage.
A failed run is not charged at all
A run that did not come back releases its hold in full. If the provider billed us anyway, that is our loss to absorb and our incident to record, not a line on your balance.
The three bands, in order: at or under the quote, you are charged the actual cost and the rest is released. Above the quote, up to 2x it, the excess is charged too. Past 2x the quote, the platform absorbs the rest and pulls the tool from the catalog immediately.
A held amount is a quote, not a cap on the charge.
What the platform takes
Provider cost, plus a markup on top of it.
One margin, applied to what the underlying call actually cost. No subscription, no per-seat fee, no minimum monthly spend, and no charge for search, inspection or reading a usage doc.
Every tool in the catalog carries the same platform markup policy: a fixed percentage of the provider's cost for that call, authored on the tool itself and applied by the pricing engine before your agent is ever quoted a number.
The exact figure for a specific call has exactly one authoritative source: omnial_execute with dry_run: true. That figure includes the markup and is exactly what gets held while the call runs. No provider is contacted to produce it. It is a quote rather than a cap; see the ceiling above.
Prices published in this catalog are the provider's own list price, deliberately: a marked-up range is not your bill either, and it would destroy the one number on the page you can check against the vendor's own rate card.
The rate is a single, uniform policy the pricing engine applies to every shipped tool, not a number this page retypes, because a second hand-typed copy of a money rule is exactly the kind of claim that goes stale silently. dry_run: true is the one place the figure is always current.
Free, always
omnial_search: finding a toolomnial_inspect: the contract and an estimateomnial_learn: the whole usage docomnial_balance,omnial_runs: your own statedry_run: true: pricing a call without making it
Charged
A successful omnial_execute, at the provider's cost for that specific call plus the platform markup. Nothing else.
Prices in this catalog are the provider's own list price, not your bill: Omnial MCP charges provider cost plus a platform markup on top, so what you are charged is higher than the figure shown. For the exact amount a specific call will cost, run omnial_execute with dry_run: true; that number includes the markup and is what we hold while the call runs. It is a quote, not a cap on the charge.
Controls
Bounds an agent cannot talk its way past.
An autonomous process spending money needs limits enforced by the system that holds the money, not by the prompt.
Per-key spend caps
Scopes
runs:execute can search, inspect and read usage docs and will be refused by omnial_execute. That refusal is stated at the moment you choose the scopes, not discovered later.The balance itself
Reconciliation
Every cent is explainable, or it is a bug.
- Append-only ledger
- Entries are never edited or deleted, and each one carries a full balance snapshot, so any balance can be re-derived from its own history.
- Per-run breakdown
- Each run shows estimated, held and actual side by side with the ledger entries that moved the money, and what Omnial MCP kept on that run as the difference between two rows you can see.
- Refunds are visible where the money is
- A refund writes a ledger reversal and touches no invoice row, so invoices and reversals are shown together and an invoice reads as refunded only when it was reversed in full.
- Usage by day
- A run's cost is attributed to the day it became real, when it settled, not the day it was submitted, so an async render that finishes after midnight does not post itself to a day you already read.
Early access: what is real and what is not
Omnial MCP's catalog is no longer empty: our first provider integration, a genuine vendor account, has been promoted all the way to live — it is browsable at /tools and callable today, the first real, working integration on this platform. Every other provider defined so far remains below live. Most are synthetic: an invented vendor with a full input and output contract, a modelled pricing structure and a hand-authored usage doc, reachable only at a reserved .example host that resolves to nothing. One entry, echoco, is not synthetic and is not a vendor either: it is an internal fixture that calls a real, free public test endpoint solely to prove the pipeline end to end, and nothing about it is for sale. Nothing about those remaining entries is a working paid integration, and no figure on this site attached to them is revenue. What that live integration proves, and what the platform around it was already built and tested for, is real: the append-only ledger, hold-before-the-call and settle-after, the pricing engine, the eight MCP tools, and now real, callable tools behind them.
Concretely, for this page: the ledger, the hold-and-settle path, the pricing engine and the top-up flow are built and tested, and a deployment with Stripe configured can take a real payment. Real, callable tools are live in the catalog today.