GST API Cost Per Verification, Line by Line

"What does a GST verification cost?" has a one-line answer on every pricing page and a different answer on every invoice. The gap is not dishonesty — it is that the advertised rate is one of six numbers that together decide what a single answered question costs you. This post lists all six, works a realistic example where two providers at the same headline rate produce bills 2.4× apart, and gives you the four levers that actually move the number.

The six numbers behind one verification

Write these down for each candidate provider before comparing anything.

A worked example: same rate, different bill

Two providers, both advertising ₹0.40 per call. You need to verify a 10,000-row vendor master, with filing history, once a quarter.

Provider A. Filing history is a separate endpoint (×2 calls). Credits expire in 90 days and you buy in annual blocks, so roughly a fifth lapses. Invalid rows are billed. 10,000 rows × 2 calls × 4 quarters = 80,000 calls at ₹0.40 = ₹32,000, plus about ₹8,000 of lapsed credits ≈ ₹40,000.

Provider B. One call returns everything. You dedupe first (10% repeats) and validate locally (2% invalid), so 8,800 real calls per quarter. 35,200 calls × ₹0.40 = ₹14,080, with no expiry loss because credits are bought as needed.

Same headline rate, 2.8× difference. Nothing in that gap came from negotiating the rate — it came from scope, hygiene and expiry. The same arithmetic across 1,000, 10,000 and 100,000 GSTINs is tabulated in what 1,000 GSTIN verifications really cost.

The engineering cost nobody puts in the spreadsheet

At low volume, the API bill is not your largest cost — your own time is. One engineer-day is worth thousands of verifications at any rate on the market, which flips the usual instinct: paying slightly more for an API whose error bodies are informative, whose rate limits are documented, and whose response shape is stable is almost always cheaper than saving ₹0.10 a call and spending three days on retries and schema drift.

Concretely, price these in: initial integration, batching and retry logic, handling 429s with backoff and jitter, monitoring, and the recurring cost of re-reading an undocumented response when a field changes. The error-handling surface is covered in the GST API error codes guide and rate limits and webhooks.

Four ways to move the number down

1. Dedupe before you spend. Normalise case and whitespace, then take distinct values. This costs nothing and is usually the single largest saving.
2. Validate locally first. The GSTIN format is a fixed 15-character structure with a check digit, so a typo can be rejected offline. Free, instant, and it removes billable calls — see GSTIN format explained.
3. Match cadence to risk. Daily re-verification of an entire vendor master is rarely justified. Verify fresh at the moments that matter — onboarding, before a large payment, before a quarter-end reconciliation — and run a cheaper background cadence otherwise, as in automated vendor GSTIN monitoring.
4. Buy into the bulk tier in one purchase. If the discount triggers on purchase size, two purchases of 1,500 credits cost more than one of 3,000 for the same volume.

The full version of this list, with the measurement method attached to each step, is in how to cut your GST verification API bill.

What one verification costs here

₹0.50 per verification pay-as-you-go; ₹0.40 once a single purchase crosses 2,500 credits. One GET /api/get-taxpayer-info/{gstin} call returns identity, status and status dates, constitution and taxpayer type, jurisdiction, places of business, e-invoicing enablement and per-period GSTR filing history with ARNs — so the "calls per useful answer" multiplier is 1, not 2. Format and check-digit failures are rejected before upstream and cost nothing. Upstream 502s are not billed. There is no subscription and no minimum, so unused allowance cannot accrue against you.

Model your own volume on the cost calculator, or see the published tiers on the pricing page and the buyer's framing on the cheapest GST API page.

Frequently asked questions