API docs

Send SMS

Send an SMS to one number at once, or to many numbers as a batch.

POST/v1/sms

Sends to one number. When the response arrives, the message has been handed to the operator.

Body

tostringrequired
An 8-digit number. +976, spaces and dashes are accepted. The supported numbers are listed below.
textstringrequired
The message. At most 4 parts.
text_by_carrierobject
A text of its own per operator: { "mobicom": "Sain baina uu?" }. That operator's numbers get it instead of text; the rest get text. Keys: mobicom, unitel, skytel, gmobile, ondo, lime.
curl -X POST https://api.wesend.mn/v1/sms \
  -H "Authorization: Bearer $WESEND_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "to": "99112233",
  "text": "Tanii zahialga batalgaajlaa. #1042"
}'
Response200 OK
{
  "id": "0b9c1a52-6f0e-4d8b-9a3c-2f1e5d7c8a90",
  "status": "sent",
  "channel": "sms",
  "to": "99112233",
  "parts": 1,
  "rate_multiplier": 1,
  "credits": 80,
  "batch_id": null,
  "error": null,
  "sandbox": false,
  "created_at": "2026-10-06T08:15:30.000Z",
  "sent_at": "2026-10-06T08:15:31.000Z",
  "encoding": "latin",
  "non_latin_characters": [],
  "balance": 4920
}

encoding says how the message was billed: latin holds 155 characters a part, unicode holds 65. non_latin_characters lists each character that made it unicode, once, up to 50. When a message written in Latin letters comes back unicode, this is why. See Which characters are Latin, below. Sending Une 10,000₮ 🙂, for instance, answers:

Response200 OK
{
  "id": "0b9c1a52-6f0e-4d8b-9a3c-2f1e5d7c8a90",
  "status": "sent",
  "channel": "sms",
  "to": "99112233",
  "parts": 1,
  "rate_multiplier": 1,
  "credits": 80,
  "batch_id": null,
  "error": null,
  "sandbox": false,
  "created_at": "2026-10-06T08:15:30.000Z",
  "sent_at": "2026-10-06T08:15:31.000Z",
  "encoding": "unicode",
  "non_latin_characters": [
    "₮",
    "🙂"
  ],
  "balance": 4920
}

If the operator does not accept the message, the answer is 502 provider_failed and the credits are refunded. The error's id is the failed message.

Messages to ONDO and Lime numbers, and messages another operator did not accept, go over a different route. That route carries no links, so any link in the text (https://…, www.…, example.mn/…) is removed before delivery. A message that is nothing but a link is not sent and fails with 502 provider_failed.

Supported numbers

The operator is read from the start of the number. An 8-digit number starting with one of these is accepted.

OperatorNumber starts with
Mobicom85, 94, 95, 99, and 5050-5099, 5550-5599
Skytel69, 90, 91, 92, 96
Unitel80, 86, 88, 89, and 5000-5049
G-Mobile78, 83, 87, 93, 97, 98
ONDO60, 66, and 8181
Lime72
Toki5500-5549 (on the Unitel network)
KAOS505 (on the Mobicom network)

A number with any other start (a landline or a foreign number, for instance) answers 422 unsupported_number. No credits are taken.

Parts and price

A long message counts as several parts. Price = parts × credits per part × holiday multiplier.

ModeOne partWhen
Latin (ASCII)155 charactersEvery character of the message is in the Latin list below
Non-Latin (Unicode)65 charactersEven one character is outside that list

One character is enough. If a message written in Latin letters holds a single emoji or ₮, the whole message, not just that character, switches to the non-Latin mode and each part holds 65 characters. A 120-character Latin message is 1 part; add one emoji and it becomes 2 parts, at twice the price.

A message longer than 4 parts answers 422 message_too_long. The error carries encoding and non_latin_characters, so the characters that made it non-Latin are named. An emoji counts as 2 characters.

Which characters are Latin

The Latin mode allows only the ASCII characters of the operators' basic alphabet. A message written in Latin letters is not always in the Latin mode: compare the two lists below.

Allowed in the Latin modeCharacters
English lettersA - Z, a - z
Digits0 - 9
Space and line breakAn ordinary space, Enter
Punctuation and symbols! " # $ % & ' ( ) * + , - . / : ; < = > ? @ _
Ends the Latin modeCharacters
Cyrillic lettersА - Я, а - я, Ө, Ү and the rest
Emoji😊 👍 ❤️ 🔥 and every other
Currency and special signs₮ € £ № ° © •
What phones and Word put in by themselvesCurly quotes ‘ ’ “ ”, « », long dashes –, the ellipsis …, a non-breaking space
Accented lettersé ü ö ñ and the rest
Nine ASCII characters[ \ ] ^ { | } ~ and the backtick. Operators bill these double, so WeSend counts them as non-Latin
MessageMode
Une 10,000TLatin, 155 characters a part
Une 10,000₮Non-Latin (₮), 65 characters a part
It's readyLatin, a straight quote '
It’s readyNon-Latin, a curly quote ’
Bayarlalaa :)Latin
Bayarlalaa 🙂Non-Latin, an emoji
  • Write an amount with T or MNT instead of ₮, and a smile as :) instead of an emoji.
  • Check text copied from Word, Google Docs or a phone: they swap a plain ' " - for curly quotes and long dashes by themselves.
  • The response's parts says how many parts the message took, and encoding with non_latin_characters says why. A sandbox key checks it without spending credits. The dashboard's Send SMS page points out the non-Latin characters and fixes them in one click.

A different text per operator

Mobicom does not deliver an SMS of 2 or more parts. So a long message to a Mobicom number is cut between words into separate one-part SMS, sent one after another. Each counts as one part, and parts in the answer is how many there are. To send fewer, give Mobicom a short Latin text while the other operators get the Cyrillic one. Each number is priced by the text it gets.

curl -X POST https://api.wesend.mn/v1/sms/bulk \
  -H "Authorization: Bearer $WESEND_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "to": [
    "99112233",
    "88112233"
  ],
  "text": "Сайн байна уу? Баярлалаа",
  "text_by_carrier": {
    "mobicom": "Sain baina uu? Bayarlalaa"
  }
}'

List the operators

GET/v1/carriers

The operators WeSend sends to and how their numbers start. Each id is a key of text_by_carrier.

curl https://api.wesend.mn/v1/carriers \
  -H "Authorization: Bearer $WESEND_API_KEY"
Response200 OK
{
  "data": [
    {
      "id": "mobicom",
      "name": "Mobicom",
      "includes": [
        {
          "name": "KAOS",
          "prefixes": [
            "505"
          ]
        }
      ],
      "multipart": false,
      "links": true,
      "prefixes": [
        "85",
        "94",
        "95",
        "99",
        "5050-5099",
        "5550-5599"
      ]
    },
    {
      "id": "unitel",
      "name": "Unitel",
      "includes": [
        {
          "name": "Toki",
          "prefixes": [
            "5500-5549"
          ]
        }
      ],
      "multipart": true,
      "links": true,
      "prefixes": [
        "80",
        "86",
        "88",
        "89",
        "5000-5049",
        "5500-5549"
      ]
    },
    {
      "id": "ondo",
      "name": "ONDO",
      "includes": [],
      "multipart": true,
      "links": false,
      "prefixes": [
        "60",
        "66",
        "8181"
      ]
    }
  ]
}
FieldMeaning
includesBrands on its network: KAOS runs on Mobicom, Toki on Unitel.
multipartfalse when it does not deliver an SMS of 2 or more parts: a long message goes as separate SMS cut between words (Mobicom).
linksfalse when links are always removed before delivery (ONDO, Lime).

Send to many numbers

POST/v1/sms/bulk

Sends one text to up to 1,000 numbers. The messages are queued and sent in the background.

Body

tostring[]required
1 to 1,000 numbers. Duplicates are merged.
textstringrequired
The text every recipient receives.
text_by_carrierobject
A text of its own per operator: { "mobicom": "Sain baina uu?" }. That operator's numbers get it instead of text; the rest get text. Keys: mobicom, unitel, skytel, gmobile, ondo, lime.
curl -X POST https://api.wesend.mn/v1/sms/bulk \
  -H "Authorization: Bearer $WESEND_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "to": [
    "99112233",
    "88112233",
    "95112233"
  ],
  "text": "Shine baraa irlee! shop.mn"
}'
Response202 Accepted
{
  "id": "5d2a7c1e-93b4-4f6a-8e0d-b1c2d3e4f5a6",
  "status": "processing",
  "channel": "sms",
  "total": 3,
  "sent": 0,
  "failed": 0,
  "credits": 240,
  "sandbox": false,
  "created_at": "2026-10-06T08:15:30.000Z",
  "completed_at": null,
  "encoding": "latin",
  "non_latin_characters": [],
  "balance": 4680
}

encoding and non_latin_characters are those of text. With text_by_carrier, each operator's text has its own in encoding_by_carrier: { "mobicom": { "encoding": "latin", "non_latin_characters": [] } }.

  • The whole batch is charged up front. Credits for each message that is not sent are refunded.
  • If any number is invalid or unsupported, the whole request answers 400 invalid_request and lists the bad entries (up to 50) in the error's invalid field. Nothing is sent or charged.
  • Follow progress with GET /batches/{id}.