How many characters an SMS has: GSM7, UCS-2, accents, emojis and segments explained

160 characters is the number everyone knows, but a single accent can bring it down to 70. Here is how an SMS is encoded, what a segment is and how to write messages that do not cost twice as much.

Where the 160 characters come from

SMS was born in the 1990s with a limit of 140 bytes per message. With the GSM7 alphabet, which uses 7 bits per character, those 140 bytes hold 160 characters. That alphabet covers the English letters, digits, basic punctuation and a few symbols, but not every character used in Spanish.

When the message contains a character GSM7 cannot represent, the handset and the network switch to UCS-2, which uses 16 bits per character: the same 140 bytes now hold only 70 characters. A single character outside the alphabet is enough to encode the entire message in UCS-2.

Which characters fit in GSM7 and which do not

They fit: letters A to Z in upper and lower case, digits, the space, the signs . , ; : ! ? ‘ ” ( ) @ # $ % & * + – / = < > and the line break. From Spanish, the uppercase Ñ fits, some acute-accented vowels do (é, for example) and so does ü, but the lowercase ñ and several accented vowels do not, and characters such as the euro sign, curly braces or the backslash take two positions each.

They do not fit: the lowercase ñ, several accented vowels (á, í, ó, ú in practice on most platforms), the ¿ and ¡ signs in some cases, emojis and any symbol outside the basic set. Because the exact rules have edge cases, the safest approach is to trust the platform counter.

Segments and concatenation

When a message exceeds the limit, it is split into segments that the handset reassembles. To make that possible, each segment reserves a few header bytes, so the capacity per part drops: 153 characters in GSM7 and 67 in UCS-2. A 200-character text in GSM7 takes two segments; the same text with a lowercase ñ takes three.

Every segment is billed as a message. That is why the num_segments field in the API response, or the counter in the console, is the single most important number for controlling cost.

How the Grupo Tecnophone platform handles it

Encoding is determined automatically from the content and the connection settings. If only GSM7 is enabled, unsupported characters are normalised or replaced with compatible equivalents (ñ becomes n, accented vowels lose the accent), which guarantees a 160-character segment. If UCS-2 is enabled, the platform uses GSM7 wherever it can and falls back to UCS-2 only when the text demands it.

Concatenation is also configured per connection: if it is disabled, the message is capped at a single segment and any surplus text is not sent. The API response reports the encoding applied (encoding), the character count (num_chars) and the segments (num_segments) for each message.

Practical recommendations

• Write without accents or lowercase ñ when the message is operational (OTP, notices, collections); the customer reads it just the same.
• Save emojis for promotional messages where the impact justifies the cost, and always check the counter.
• Use the console preview or test in sandbox to see the segments before sending.
• Cut the filler: long greetings, signatures, repetition. One fact and one action fit in 160 characters.
• If you genuinely need more room, consider a short link to a page with the detail.

Frequently asked questions