Why UnifAPI prices public-data reads at $0.001 credits with published per-operation minimums, so Skills scale by usage, not subscription tiers.

In short: This is the pricing pillar: UnifAPI prices returned records so agent operators can estimate workflow cost without committing to subscription tiers.

Most API marketplaces price like this: pick a tier (Basic / Pro / Ultra), pay a fixed monthly fee, get N requests included, hit a hard cap, then either upgrade or get cut off. RapidAPI is the most familiar example. It's also the most painful.

We considered doing the same thing. The financial argument for tiered subscriptions is real — it's predictable MRR, it's the dominant SaaS pattern, and it's easier to fundraise against. We threw it out anyway.

The problem with caps

Tiered plans punish the exact moment that should be celebrated: your product taking off. The day your traffic spikes is the day your provider cuts you off — or charges you 5x the per-call rate for overage. That asymmetry kills experiments and starves growth.

The cap also forces a planning fiction. You don't know how many records you'll need this month, but you have to pick a tier as if you did. The result: you either overpay for headroom or underpay and get throttled.

What we ship instead

First top-up bonus — top up a prepaid balance, and every workspace's first top-up includes a one-time 2,000-credit bonus. Live APIs unlocked, the MCP server included. You can browse Skills before registering; OAuth happens when the agent needs tool calls.

Pay-as-you-go — top up a balance whenever you want. Credits cost $0.001 each, with per-operation minimums published from the OpenAPI billing contract. No monthly fee. No hard cap. Pull a single standard record this month or a million — the bill matches.

Enterprise — for teams who need an SLA, procurement-friendly contracts, or volume discounts on the per-record rate. Sales-led, contract-based.

Why per record, not per call

A call is a bad unit. One call to a creator profile endpoint returns one record; one call to a social search endpoint can return fifty posts. Per-call pricing means a small request and a big request cost the same — which forces the provider to price the worst case, which means the small request is overpaying.

Records are what the agent actually consumes. The work the model has to read and reason about is the row of data, not the HTTP envelope. Pricing per record lines up the bill with what you're actually using, and makes the unit easy to estimate before you ever make a call.

What this means for you

Budget by usage, not tier. The hardest part of agent infrastructure is knowing how much a workflow will consume. With per-record pricing, you can run the Skill, inspect the record count, and decide whether to deepen the analysis.

The model bill stays with the agent product you already use. UnifAPI bills the public-data layer only, which keeps the workflow easy to reason about before it becomes production code.

What to read next

Pricing - the live pricing page for credits and top-ups.

UnifAPI vs RapidAPI - applies the pricing argument to a marketplace comparison.

Build a Reddit research agent - shows record pricing on a concrete workflow.