Every check MailSift runs, and what each one returns.

Use MailSift when you need more than a regex and less hand-holding than an enterprise implementation project. It is shaped for developers, revops teams, and operators who need fast answers at batch scale.

15+ checks on every request

Each call runs the whole pipeline, well past syntax and MX. The right-hand column is the field your code reads.

CheckWhat it doesField in the response
Syntax & normalizationRFC-compliant local part and domain checks plus provider-aware normalization for Gmail-style aliases."syntax": true
MX lookupPull live MX records and ignore null-MX domains so undeliverable hosts never enter your funnel."mx": true
Disposable detectionCatch throwaway addresses from 500k+ known providers. MX heuristics detect new ones automatically."disposable": false
Relay & forwardingUnmask SimpleLogin, AnonAddy, Apple Hide My Email, and Firefox Relay that hide the real sender."relay": false
Domain intelligenceCheck domain age, MX provider, and whether the domain is routable, disposable, or parked."domain_age_days": 4823
Typo correctionSuggest corrections for common domain typos like gmial.com — before they turn into hard bounces."did_you_mean": "gmail.com"
SPF & DMARCValidate SPF and DMARC records to confirm the domain has proper email security configuration."spf": true, "dmarc": true
Role-based detectionFlag info@, sales@, and other inboxes that route to teams instead of a single human."role": false
Free provider checkIdentify gmail, yahoo, outlook, and other free providers for clean segmentation downstream."free": true
Catch-all detectionSpot domains that accept every address so you can route catch-all hits to a separate workflow."catch_all": false
MX provider attributionMap every domain to its hosting provider — Google, Microsoft, Proofpoint, Mimecast, and more."mx_provider": "Google"
Local-part anomaliesDetect synthetic patterns like long-digit prefixes that often slip past basic regex validators."score": 82
Trust scoringAll 15+ signals combine into a single 0-100 score. Route by risk, not guesswork. Tune per environment."score": 95
Normalized emailCanonical address form for deduplication — strips plus-aliases, dots, and provider-specific noise."normalized_email": "[email protected]"

Two modes on one credit balance

Standard is the default on every endpoint. Deep adds one thing: an SMTP conversation with the recipient’s mail server about the mailbox itself. Same endpoints, same credit balance, one parameter apart.

Default

Standard

mode=fast
1credit per email

Everything MailSift can establish from DNS, reputation data, and a domain-level probe.

Speed

Under 100 ms on average.

Answers

Every check in the pipeline above, plus the 0–100 trust score.

Won’t answer

Whether your recipient’s mailbox exists. Standard can prove an address is unreachable. It never proves a mailbox is real.

Pick it when

Signup and checkout validation, inline field checks, list hygiene, routine bulk cleaning.

Opt in per request

Deep Verification

mode=deep
up to5credits per email

Everything in Standard, plus an SMTP conversation with the receiving mail server about the mailbox itself.

Speed

Slower. It waits on a third-party mail server, so no completion time is promised.

Answers

Mailbox acceptance, recipient-specific rejection, catch-all resolution, greylisting detection, and provider-specific interpretation.

Won’t answer

Mailbox existence at providers that conceal it. Those come back as unknown, not as a guess.

Pick it when

Cold outreach, re-engagement sends, high-value signups, anything where a bounce is expensive.

How Deep is billed

Deep costs 5 credits only when the SMTP conversation produced real mailbox- or domain-level intelligence. If the standard checks already settled the address, or the conversation was defeated by a timeout, throttling, or a policy block, it costs 1 credit or less. You are never charged 1 plus 5, and there is no separate Deep subscription.

Deep requested, standard checks settled it first:"mode": "deep", "credits_used": 1

Deep reports what the server actually revealed

Some providers will not confirm mailbox existence. Yahoo and AOL accept recipients they cannot vouch for. Secure gateways like Proofpoint, Mimecast and Barracuda accept every recipient at the perimeter. For those, MailSift returns unknown, usually with catch_all true, instead of dressing an acceptance up as proof. The raw result stays in the response so you can route on it yourself.

Server answeredgmail.com
"smtp": {
  "status": "deliverable",
  "response_code": 250,
  "catch_all": false,
  "mx_host": "gmail-smtp-in.l.google.com",
  "provider": "google"
}
Provider concealed ityahoo.com
"smtp": {
  "status": "unknown",
  "sub_reason": "provider_unverifiable",
  "response_code": 250,
  "catch_all": true,
  "provider": "yahoo"
}

Same request, same response shape. The second address is not an error and not a failed check. It is a mailbox the provider will not discuss, reported as exactly that.

Full smtp field and status reference

Three ways to send addresses in

Call the API inline, run a batch, or listen for events. Pick the entry point that fits the job.

Real-time API

A single GET request returns disposable status, domain reputation, risk score, and 15+ checks. Sub-second median response times in every region.

  • GET /verify with one URL parameter
  • Per-API-key rate limits, predictable quotas
  • Cached results for repeat hits inside a window
  • JSON response includes full check breakdown

Bulk CSV

Upload a CSV, get a job id back instantly, and stream progress over SSE. Read parsed rows for in-app filtering or download the finalized output.

  • Multipart upload up to 32MB per job
  • SSE progress stream for live dashboards
  • Paginated parsed rows for filtering
  • Final CSV download when the job completes

Webhooks

Signed webhooks deliver bulk job events to your endpoints so you can wire them into a CRM, queue, or warehouse without polling.

  • HMAC-SHA256 signature on every payload
  • Webhook secret available from /account
  • Fired on bulk job completion and failure
  • Stable event schema versioned with the API

The same request in four languages

It is one GET with an API key header. The request shape does not change with the language you write it in.

Read the API reference
curl "https://mailsift.dev/api/v1/[email protected]" \
  -H "X-API-Key: msk_your_api_key_here"

Run your first verification in under five minutes

Sign up, generate a key, and send one address. Credits never expire.