Pricing an API platform is a different problem from pricing most SaaS. The unit of value isn't a seat, it's a request — and requests are cheap individually but the relationship between price and usage has to feel obviously fair, or developers churn the moment they hit a limit.
The tiers
- Free — $0/month, 500 calls/month, 10 req/sec, community support.
- Starter — $19/month (₹1,549), 10,000 calls/month, 30 req/sec, 48h email support.
- Growth — $49/month (₹3,999), 100,000 calls/month, 100 req/sec, includes /rag, 24h email support.
- Scale — $199/month (₹16,499), 1,000,000 calls/month, 500 req/sec, priority support.
- Enterprise — custom pricing, unlimited calls, custom rate limit, dedicated Slack support.
Each tier is roughly a 10x jump in both quota and price from the one below it. That ratio wasn't arbitrary — it's the simplest mental model for a developer to reason about: “if I'm close to my limit, the next tier gives me 10x the room for roughly 2.5–4x the price.” The quota-to-price ratio actually improves as you go up, which is the right direction — your highest-volume customers should get the best per-call economics, not the worst.
Why Starter moved from $9 to $19
Starter originally launched at $9/month. We moved it to $19 after checking where single-purpose competitors actually price: AbstractAPI's entry paid tier runs around $17/month, APILayer's paid tiers range roughly $15–60/month depending on the specific API. At $9, we were underpricing a platform that covers 34 namespaces against competitors who cover one. That's not a sustainable position — being meaningfully cheaper than a single-purpose competitor while offering far more breadth undervalues what one key, one bill, and one dashboard across every namespace is actually worth.
Why a flat monthly quota instead of metered, pay-per-call billing
Metered billing (pay exactly for what you use, no tiers) sounds fairer in the abstract, but it's worse for a developer trying to budget. A flat monthly quota means you know your maximum bill before you write a single line of code. Metered billing means your bill is a function of how successful your product is — which is exactly the moment you don't want a surprise invoice. Quota-based tiers also make our own infrastructure costs predictable, since rate limiting (not billing logic) is what actually protects against abuse.
Why /rag specifically gates at Growth, not Free
Every other namespace is available on every plan, including Free. /rag is the one exception, because it has a real per-call cost we don't control — every document upload and every query makes an OpenAI API call on our end. Free-tier abuse of an LLM-backed feature is a fundamentally different risk than free-tier abuse of a pure-computation endpoint like an EMI calculator, so it's gated to the tier where the unit economics actually work.