GSTIN API Pricing: How to Compare Quotes That Are Not Comparable

You have three quotes for a GST verification API open in three tabs. One is priced per call. One sells credits in packs, the per-credit rate falling as the pack grows. One is a monthly subscription with an allowance baked in and an overage rate in the footnotes. Comparing the three biggest numbers on the three pages is the quickest way to answer "which is cheapest", and the most reliable way to pick the wrong one.

The short answer: you cannot compare GSTIN API pricing by rate, because each vendor charges its rate against a different unit. You compare by modelling your own twelve months of volume — retries, duplicates and the quiet months included — through each pricing structure, then reducing all three to one number: cost per verified vendor.

Why there is no price table on this page

Because it would be wrong, in a way that is worse than useless. GST data pricing in India moves constantly: vendors reprice, add tiers, change what counts as a billable call and revise free allowances, often several times a year. A page listing named vendors and their rates is accurate for a few weeks and then quietly misleads everyone who finds it afterwards — which, given how search works, is most of its readers.

What does not move is the structure. The five models below have been the shape of this market for years, and the hidden line items in the next section have been catching buyers out for just as long. Learn the structure once and you can price any quote yourself against whatever numbers are on the vendor's page this morning. That does not go stale.

The five pricing models you will be quoted

A pricing model is the unit a vendor charges against and the rule that turns your usage into a bill. Five are in common use for GST verification, and the material difference between them is not the rate — it is how each behaves when your volume does something you did not forecast.

The right model depends on the shape of your usage, not on the vendor. A company verifying a steady four thousand GSTINs a month and a company doing one ten-thousand-row cleanup a year should not pick the same structure, and the lowest advertised rate is frequently wrong for both.

The pattern worth flagging: fixed-fee models transfer forecasting risk from the vendor to you, and that risk is what you buy the discount with. Good forecast, you win. Lumpy volume — and verification volume is lumpier than people expect, because onboarding drives it — and you are paying a premium for a discount you never collect. The same dynamic runs through every subscription a business accumulates, which is why it pays to audit what you already subscribe to before adding a line.

The line items that actually decide the bill

The rate is on the pricing page. These live in the documentation, the fair-use policy, or nowhere at all until you ask — and each can move your effective cost by more than the gap between two vendors' headline rates.

Are failed calls billed?

The highest-leverage question. When a GSTIN does not exist, is malformed, or the upstream errors out, does that decrement your balance? Policies genuinely differ: some bill every request, some only successful lookups, some everything except upstream failures. On a clean production stream it barely matters. On a first pass over an inherited vendor master, where a real slice of rows are junk, it shifts the total materially.

What happens to repeated lookups?

If you call the same GSTIN twice in a day, is that two billable calls? Some providers serve a short-lived cached response free or reduced; others charge full freight. It interacts with your own architecture too — if you are not deduplicating before you call, you are paying for your own bug.

Is every endpoint priced the same?

Often not. A registration lookup and a filing-status call cost the provider differently to serve, and filing history is usually the pricier one. It is also the field with the most value in it for anyone worried about input tax credit exposure. If your workflow leans on return filing status, price the quote against that endpoint, not the cheapest one on the page.

Is there a minimum commitment or expiry?

Two versions of one trap. A monthly minimum means quiet months cost what busy ones do. Credit expiry means the pack you bought for a cleanup that slipped two quarters is now worth nothing. Neither is hidden maliciously; both get missed because buyers read the rate and skim the terms.

What is the overage rate, and is it a cliff?

In tiered models, ask what you pay per call beyond the allowance, and whether exceeding it bumps you into the next tier wholesale. The second is far more expensive and far less commonly asked about.

Is there annual lock-in, and what does it buy?

Annual prepayment usually earns a discount, sometimes nothing but a longer exit. A twelve-month commitment in a market where rates drift downward is a bet that the discount beats the drift. Make it knowingly.

Computing your true cost per verified vendor

Cost per verified vendor is your total annual spend divided by the number of distinct businesses you successfully verified that year. It is not cost per call, and the ratio between the two differs by provider — which is exactly why comparing advertised per-call rates does not work.

Three things drive a wedge between calls and verified vendors:

• Duplicates. The same GSTIN appears repeatedly across a vendor master, an invoice stream and a re-check schedule. Whether those bill depends on the vendor's cache policy and on whether you deduplicate first.
• Retries. Timeouts, transient upstream failures and your own retry logic generate calls that produce no new information. A provider with worse availability generates more of them, so its effective cost is higher than its rate suggests.
• Non-verifiable rows. Malformed numbers, cancelled registrations, entries that were never GSTINs. They consume calls and return no verified vendor — unless the provider does not bill them, which is the first question above.

The method is five steps and fits in a spreadsheet in under an hour:

A worked example with illustrative numbers

Everything in this section is made up. The rates below are round placeholder figures chosen to make the arithmetic visible — not any vendor's prices, not market averages, not an estimate of what you will pay. Replace every number with figures you collect today.

Take a mid-sized distributor. The vendor master holds V = 2,000 distinct GSTINs. Policy is a monthly status re-check on active vendors, so E = 24,000 verification events a year. Logs show retries, escaped duplicates and unverifiable rows add roughly 15%, so C = 27,600 billable calls. Price that through three imaginary structures, using a made-up base unit of ₹1.00 per call purely so the comparison reads clearly:

Per verified vendor at the forecast: A is ₹13.80, B is ₹12.00, C is ₹12.00. Two tied, and a spread small enough that it should not decide anything on its own.

The stress columns are where the exercise earns its keep. B ties for best at the forecast and is comfortably worst at triple volume, because the overage rate sits above the effective in-tier rate — it wins nothing and can lose badly. C is cheapest at triple volume and burns more than half of what you bought if volume halves. A is never cheapest, never surprises you, and is the only one that costs less when you use less. Predictable volume favours the committed structures; lumpy onboarding makes the boring one cheaper than it looks.

What to ask that is not about price

Once two quotes land within a few percent of each other — which, in a competitive market, they usually do — price stops being the decision and these start.

• Is the uptime commitment a number or a sentiment? "Highly available" is marketing. A stated percentage, a measurement window and a remedy when it is missed is a commitment. Ask which you are being offered.
• How fresh is the data? Some responses are live, some come from a cache of varying age. For a legal name that is irrelevant. For registration status — the field that changes and costs you money when it does — a stale answer is a wrong answer delivered confidently.
• What happens when the upstream portal is down? GST infrastructure has outages and no vendor controls that. They do control the behaviour: a clear error you can retry, a queue that completes later, or a silent stale response. Establish which, and whether those calls are billed.
• What is the support commitment, and on what channel? The scenario that matters is not a sales question on a Tuesday. It is a bulk job failing at 6pm before a payment run. Ask who answers then.
• Are there rate limits, and what happens at the ceiling? A job that trips an undocumented limit halfway leaves you with a partially verified list and no clean record of where it stopped — which matters more than the rate if bulk verification is your main use case.
• What does the exit look like? Can you export your verification history? If the audit trail lives only in their dashboard, switching vendors costs you the record of what you checked and when, which is the part an auditor asks about.

When building direct against the source makes sense

The honest version, from a vendor: sometimes it does.

Building direct is sensible when GST verification is core to what you sell rather than a check inside a process you run. If verification is the product, owning the pipeline is a reasonable strategic call and the per-call economics eventually favour it. It is also sensible at volumes high enough that the engineering cost amortises to very little per call, especially if your team already runs data pipelines with an on-call rotation — then you are adding another instance of a problem you handle, not a new class of it.

It is not sensible when verification is a feature of something else. The build is not the hard part; anyone can write the client. The hard part is everything after the demo: credential management, upstream schema changes nobody announces, retry and backoff behaviour, rate limits, monitoring that tells you the data went stale rather than that the endpoint returned 200, and somebody carrying a pager for a compliance dependency that is now yours. That maintenance runs forever and is paid in engineering time — the most expensive line item in the comparison and the one that never appears on a pricing page.

This is the general build versus buy decision in miniature: buy the commodity, build what your customers pay you for. Live GST data is a commodity; what you do with it usually is not. And if you are still deciding whether you need live lookups at all rather than a local format check, settle that first — our post on verification versus format validation covers it, and a format check is free, which is always the cheapest tier.

Common questions

Why will nobody give me a simple per-call price? Many vendors will, and a pure per-call rate is the easiest structure to reason about. The complication is that per-call pricing gives the vendor no volume certainty, so it usually carries the highest unit rate. Tiers, packs and commitments all trade your certainty for their discount, and whether that trade is good depends on how predictable your volume really is.

Is the cheapest per-call rate the cheapest option? Not reliably. The rate is multiplied by billable calls, and billable calls depend on whether failed lookups count, whether cached repeats count, and how many retries the provider's availability forces on you. A higher rate that does not bill failures can produce a lower bill on a messy vendor master. Model the total, not the rate.

How do I estimate volume before integrating anything? Count distinct GSTINs in your existing vendor master, multiply by your intended re-check cadence for annual verification events, then add a margin for retries and bad rows. If you have no basis for that margin, run a small paid pilot for a month and measure it. A measured factor from 500 real calls beats a confident guess over 50,000.

Should I commit annually to get a discount? Only after measuring your volume for a few months rather than forecasting it, and only if the discount exceeds the risk of the volume not arriving. Annual commitments work well on stable, mature usage. Signed during an integration, they are a common way to prepay for a growth curve that never shows up.

What if my volume is one big cleanup and then almost nothing? That is the classic case for prepaid credits or straight per-call billing, and the worst fit for a monthly subscription, which bills you through eleven quiet months. Check credit expiry before you buy: a cleanup that slips a quarter is the exact scenario expiry clauses are written for.

Does DL Minds publish its own pricing? Yes. gstinapi.com is our product and its current rates sit on its own pricing page, which stays accurate because it is updated when they change. We keep them off this post so the method here is still worth reading a year from now, and so you can apply it to us on the same terms as anyone else.