Free Trade Data API: Exactly What the Free Tier Includes
Every limit printed here was read out of the running service configuration at generation time - the same constants the production API enforces - alongside the October 19, 2026 USITC dataset (23,193 lines).
Most "free API" pages hide their actual limits behind a sales form. This page does the opposite: it enumerates precisely what costs nothing, what is metered, and where the honest boundaries sit, so you can architect against real constraints instead of discovering them in production.
What is free, line by line
| Capability | Free tier | Paid (metered key) |
|---|---|---|
| Read-only tariff endpoints | Unlimited within fair use, no key | Same data, accounting per call |
| Dataset coverage | All 23,193 ten-digit HTS lines (2026-10-19 USITC export) + Chapter 99 layers | Identical dataset and vintage |
| Saved watchlists | 3 items | 25 items (Pro monitoring) |
| Batch endpoint pacing | 60 requests/minute default guardrail | 60 requests/minute default guardrail |
| Cost accounting | None (anonymous use) | $0.06/call list · $0.05 via $10 credit pack |
The watchlist ceilings, batch pacing and pricing in this table were verified against the deployed service's enforced quota constants at generation time - the same numbers production applies, not marketing copy.
Why anonymous-unlimited is sustainable here
The read endpoints serve static-derived data - a snapshot joined with its remedy layers - cached at the edge, so marginal cost per anonymous call is pennies of bandwidth, not a database query storm. Metering exists for teams who need receipts: procurement wants line-item invoices, CI pipelines want hard budgets, and prepaid credits cap worst-case spend at the balance actually loaded. Both modes hit the same production base URL and the same dataset vintage; the key changes accounting, never truthfulness.
Pricing mirrors the published API reference, which documents every response field with live examples.
Fair use, stated plainly
"Unlimited" always coexists with abuse protection. The practical rules: keep concurrent connections modest, prefer the batch endpoint over request storms when walking many codes (it paces itself at 60/minute by default), and honor cache headers instead of re-fetching unchanged lines. These guardrails keep latency flat for everyone; they are enforced technically, not just aspirationally.
Designing against the free tier: three patterns
- Nightly reconciliation. Pull your watched codes once after the daily refresh, diff locally, alert on change - zero calls during peak hours, zero dollars ever.
- Hybrid pricing engine. Cache General rates locally; call the live lookup only when a Chapter 99 layer could have moved - China-origin goods or remedy-scope products.
- Budget-fenced bursts. Load the $10 credit pack once, cap exposure at $10, and let the metered ledger show exactly which endpoints consumed it - useful for proving internal demand before committing to Pro monitoring.
Beyond the API: the free surface area of the whole tool matrix
The API is one front of the free tier. The same dataset powers the interactive HS code browser across all 96 data-bearing chapters, the landed-cost calculator with its stacked Chapter 99 arithmetic, a 2,350-ruling CBP CROSS mirror, and every guide in the library - all anonymous, all ad-supported by nothing heavier than an occasional upgrade prompt. If your workflow is human-scale rather than machine-scale, start there; the API becomes worth a key when volume or automation enters the picture.
From API rows to answered questions: the remedy-screening path
A frequent integration pattern deserves its own walkthrough. Pricing teams rarely need raw JSON; they need "does this China-origin line owe more than the base rate, and by how much". The workflow that answers it with free tools only: resolve the ten-digit line in the HS code browser, confirm the Chapter 99 layer through the lookup endpoint or the landed-cost calculator, then screen product scope against the AD/CVD case lookup covering recent antidumping and countervailing orders across dozens of countries. For FTA-partner origin, the FTA eligibility checker closes the loop by testing whether the Special column's Free rate is claimable. Each tool cites its dataset vintage on-page, so a screenshot taken today carries its own expiration date.
What the paid tier adds (and what it deliberately does not)
Paid access changes accounting and workflow surface, never data truth. Pro monitoring watches your saved codes for rate changes and emails diffs; metered keys add receipts and budget caps. What no tier sells you is exclusive data - every statutory number any plan can see comes from the same public USITC pipeline, which is why the honest differentiator is automation depth (alerts, exports, batch pacing) rather than information asymmetry. That stance is deliberate: reselling public data behind a paywall invites a commodity price war this product does not run on.
Versioning and freshness, without surprises
Data that changes under a caller is worse than data that fails loudly. Every response therefore pins its dataset vintage, and the schedule refreshes on USITC's release cadence rather than ad-hoc edits; downstream systems diff the vintage field to detect refreshes deterministically instead of guessing from value drift. The same discipline runs through the whole tool matrix - tariff-line pages cite their extract date, guides recompute headline figures at generation time and refuse to publish when source data drifts, and this page itself verifies its quota figures against the deployed service configuration before it will render.
Frequently asked questions
- Is there a free API for US tariff and trade data?
- Yes - several, at different depths. The official USITC HTS REST service publishes the raw tariff schedule free; UN Comtrade offers a public preview of trade flows; and ExportSonar's own read-only endpoints (/api/v1/) are free and unlimited without signup, returning ten-digit HTS lines with General rates plus matching Chapter 99 remedy layers so one lookup yields the stacked effective rate rather than just the statutory base.
- What are the rate limits of free customs data APIs?
- ExportSonar's anonymous access has no hard request counter - it is unlimited within normal fair-use bounds, no key and no signup step. Paid keys exist only to make accounting explicit: each successful call debits USD 0.06 (list rate) or $0.05 via the $10 credit pack from prepaid credits. Fair-use guardrails protect the batch endpoint at 60 requests per minute by default. For third-party free APIs, expect per-day token quotas and attribution requirements - check each provider's current terms before designing around them.
- Which API returns HTS duties plus AD/CVD layers?
- ExportSonar's tariff lookup joins the base-schedule line with its matching Chapter 99 provisions, so China Section 301 surcharges and similar measures appear in the same response; the companion AD/CVD case database covers 211 recent antidumping and countervailing cases across 33 countries for scope-level screening. Every documented response field is exercised by the site's own front end, and each response pins the dataset vintage so downstream systems detect refreshes deterministically.
- Do free tiers require a credit card or expire after a trial?
- Not on ExportSonar: the free tier is a permanent plan, not a trial - full tool matrix, all 23,193 tariff lines, saved-watchlist space limited to 3 items versus 25 on Pro. No card is collected for anonymous API use at all; metered keys are prepay-only, so nothing silently bills after a trial window.
Sources: ExportSonar public API reference · USITC HTS REST export