Transparent AI Pricing: See the Tokens and Credits Behind Every Task
Most AI products bill you a flat subscription and quietly absorb (or pad) the real cost of each task. When usage is metered, it's usually a number you only discover at the end of the month. SaintSal went the other way: show the cost of every task, as it happens.
The problem: AI pricing you can't see
Autonomous agents have genuinely variable cost — a quick lookup is cheap, a stubborn multi-page task is not. Hiding that behind a flat fee means light users subsidize heavy ones, and nobody can predict their spend. Opaque metering is worse: a surprise bill with no per-task breakdown.
How SaintSal meters tasks
Two ideas borrowed from how payments actually work:
- Reserve, then settle. A small base cost is reserved the moment the cloud browser starts (so you can't start a task you can't afford), then the final amount is settled when it finishes — base plus a per-step charge, capped so a long run can never blow past a known ceiling.
- Refund on no-op. If a task fails before doing any real work, the reservation is refunded automatically. You pay for results, not glitches.
And critically, every finished task shows its tokens and credits used, right in the result panel.
Why this is unique in the market
Three things most AI tools don't do:
- Per-task visibility. You see exactly what a task cost — tokens and credits — instead of a mystery monthly total.
- A real cap. Variable work, fixed ceiling. No runaway tasks.
- Fairness by design. Reserve-and-settle plus auto-refunds means you pay for value delivered, not for the platform's overhead or its failures.
Why transparency is a growth feature, not just an ethics one
When people can see what something costs, they trust it and use it more. Transparent metering turns pricing from a source of anxiety into a reason to choose you — especially against competitors whose costs are a black box.
See how it works on the Web Assistant, and check the pricing page for the credit tiers.
Frequently asked questions
How does SaintSal price AI tasks?
Usage-based credits. A web-assistant task has a small base cost reserved when the cloud browser starts, plus a per-step charge settled when it finishes, capped so a long task can never surprise you. You see the tokens and credits each task used.
What is reserve-and-settle billing?
Like a card pre-authorization: a base amount is held when the task actually begins, then the final cost is settled at the end based on real work done. If the work is lighter than the cap, you pay less.
Do I get charged if a task fails before doing anything?
No. If a task fails before any real work happens, the reserved cost is automatically refunded. You only pay when the agent actually did something.
Can I see how many tokens a task used?
Yes. Every completed task shows the tokens consumed and the credits charged, right in the results — so cost is visible per task, not hidden in a monthly bill.
Put an AI agent to work
Build and run autonomous agents — a cloud-browser web assistant, an app builder, and an agent API.
Try the Web AssistantKeep reading
Why a Good AI Agent Asks Before It Acts
Most agents guess and return mush. SaintSal asks 1-2 sharp questions only when it changes the result — then delivers something concrete. Why that matters.
Cloud Browser vs. Browser Extension for AI Web Agents
Most AI browser agents take over your local Chrome via an extension. SaintSal runs a real browser in the cloud instead — here is why that matters.