Unicode Emoji 18.0: 3,972 Emoji and 19 New Additions

Unicode shipped Emoji 18.0 in September 2026, and this emoji 18.0 explained guide covers what changed, how many emoji exist now, and how keyword matching turns a typed word into a pictograph. The release brought the fully qualified RGI set to 3,972 emoji and added nineteen emoji: nine standalone characters plus skin-tone variants for two thumb signs. It is a small release, but it matters if you build or use a text-to-emoji converter, because the keyword dictionary behind that tool comes from CLDR 48.2, dated 2026-03-17.

Two sources disagree on the exact ship date, and both appear below rather than picking one. UTS #51's version table lists Emoji 18.0 at 2026-09-15, while Unicode's blog, its releases page, and Emojipedia date it 2026-09-16. The rest of this explainer uses the sourced numbers and flags where they conflict.

What's New in Emoji 18.0

Emoji 18.0 added nine standalone characters: Cracking Face, Leftwards Thumb Sign, Rightwards Thumb Sign, Monarch Butterfly, Pickle, Lighthouse, Meteor, Eraser, and Net with Handle. It also added five skin-tone variants for each of the two thumb signs, which brings the count of genuinely new selectable emoji to nineteen. After the release, the full RGI set stands at 3,972 emoji, including every skin-tone and gender variation.

The category breakdown shows where the set lives. People & Body holds 2,430 emoji, by far the largest group. Flags account for 270, Objects 268, Smileys & Emotion 172, Travel & Places 221, Animals & Nature 161, Food & Drink 132, Symbols 224, Activities 85, and Components 9. For context, Unicode 6.0 in 2010 encoded 719 emoji per the Unicode Emoji Counts page; a commonly cited figure of 722 does not match that sourced page and is treated here as unverified.

Emoji 18.0 Explained: The Release Date

When was Emoji 18.0 actually released? The honest answer is that two dates circulate, and neither source yields to the other. UTS #51's own version table lists 2026-09-15. Unicode's releases page and Emojipedia say 2026-09-16, a Wednesday. Until Unicode reconciles its own table, attribute both dates and treat the day as unresolved rather than repeating one as settled. The version number itself is not in dispute: this is Emoji 18.0, part of Unicode Standard 18.0, and UTS #51 was revised to rev 31 on 2026-08-28.

How a Word Becomes an Emoji

A text-to-emoji converter does not guess pictures. It matches words against the CLDR keyword table, where every emoji carries a short name and a pipe-separated list of keywords. CLDR 48.2 (2026-03-17) is the dictionary behind ToolSura's converter; CLDR 49 was still in beta as of 2026-09-25, with a 2026-10-22 release date listed. The matching flow is short:

1. The user types a word such as "dog."
2. The matcher searches CLDR annotations for that keyword.
3. A match resolves to a code point, which the page displays as an emoji.
4. Nothing is uploaded; the match runs on the user's device.

Keywords are not unique. Several emoji can share a word, and the list is drawn from associations, not a formal taxonomy, so the same input can point to more than one pictograph.

Why Some Words Don't Convert

A word converts only when it appears in an emoji's CLDR short name or keyword list. Type "crying" and nothing happens if CLDR labels that face "tears of joy"; the synonym is absent. Platforms also implement subsets differently, so a keyboard may match only the unique name and ignore the full CLDR keyword list. CLDR strips U+FE0F (variation selector-16) from stored code points, and applications must handle that selector themselves. Keep about five to six keywords per emoji, and treat the list as a guide, not a dictionary of every synonym.

Byte Costs, Flags, and JavaScript Length

Most emoji live in the supplementary plane and cost four bytes in UTF-8, but that blanket claim needs nuance: only about 30.1 percent of emoji fit in four bytes, with a median of eight bytes and a mean of 11.33. Flag emoji cost about eight bytes because each flag is two regional-indicator symbols, each four bytes. ZWJ sequences cost still more: a family emoji (man, woman, girl, and boy joined by zero-width joiners) runs about 25 bytes, and a two-skin-tone kiss reaches 35 bytes.

JavaScript counts UTF-16 code units, not characters. "😄".length is 2, while ["..."😄"].length is 1, because a supplementary-plane code point splits into a surrogate pair. A flag such as 🇺🇳 uses two regional indicators, so it spans four UTF-16 code units. Use codePointAt() when you need surrogate-pair-aware indexing; charCodeAt() is not.

Emoji Is a Character, Not a Font

An emoji is a Unicode code point, not a typeface. The same code point can look different on an iPhone, an Android device, and a desktop browser because each vendor draws its own artwork, and Unicode does not specify the visual representation. A flag pair may even display differently across platforms. ToolSura outputs plain Unicode text, so what you copy is portable: it pastes, searches, and announces the same on any device. Only the rendering font changes.

Screen Readers and WCAG 2.2 H86

Screen readers announce an emoji using its pre-defined name, which may not match what you meant. A right-pointing triangle is read as "black right-pointing pointer," not "play," so an emoji used as a button label misleads users. WCAG 2.2 technique H86 covers text alternatives for emoji under SC 1.1.1. The reliable fix is to wrap the emoji in an element with role="img" and an aria-label that states the intent; the alt attribute does not work on a span. Multiple emoji in a row are also hard to disambiguate by ear, so use them sparingly in functional text.

Related Tools

If you want to convert words yourself, try the text-to-emoji converter, which runs client-side with no upload and draws from the same 3,972-emoji pool. For the encoding side of emoji and other text, the UTF-8 / Unicode converter shows byte costs and code points side by side. These pages are tools, not explainers, so they complement rather than duplicate this post.