Telegram font generator: 17 styles you can paste

Every “font generator” for Telegram works the same way, and almost none of them say what it is doing. Nothing here installs a typeface. The tool swaps each of your letters for a different character that Unicode already defines with a different shape: the A you typed becomes U+1D400 MATHEMATICAL BOLD CAPITAL A, a distinct letter that lives in a block Unicode encoded so a physics paper could tell a bold vector 𝐇 from a plain H.

That distinction is the whole story. Because the shape is baked into the character rather than applied to it, the text keeps its look everywhere it is pasted — including the Telegram fields that accept no formatting at all. And because it is a different character, everything that works by matching characters stops working.

Where these actually belong

Telegram has two kinds of text field, and the split decides which approach is right.

Field	Native formatting	Unicode characters
Message and caption text	Yes	Works, but wasteful
Display name, bio	No	The reason this page exists
Channel or group title	No	Works
Username	No	Rejected — Latin letters, digits and underscores only

A bio is the clearest case. Telegram allows 70 characters there for a regular account and 140 for Premium, offers no bold, and a single small-caps or circled word breaks up an otherwise flat block. A channel title is the second: it is the line a subscriber scans past forty other chats, and one styled word does the work that a bold tag would do if there were one.

Where they do not belong

Inside a post. Telegram’s own message formatting covers bold, italic, underline, strikethrough, spoiler, inline code, code blocks, block quotes, expandable quotes, links and custom emoji — all as real entities attached to plain letters. That is not a lesser version of this trick; it is the better one everywhere it is available. Native formatting is searchable, screen readers read it as ordinary words, and it costs one character per character.

If typing that markup by hand is the reason you are here, that is a solved problem: the rich post editor composes headings, tables, lists, quotes and buttons visually and sends them as Telegram’s own rich message, and how it works in practice covers what renders where.

The three costs

Search. Telegram matches characters, not shapes. Rename a channel to fraktur and it stops appearing for anyone who types the name normally. A styled accent inside an otherwise plain title keeps you findable; a fully styled title does not.

Screen readers. Assistive software announces these as mathematical symbols, spells them letter by letter, or skips them. Anything a reader has to understand — what your channel is about, what to do next — stays plain.

Length. The smallest of the three, and worth stating precisely, because almost every page on this subject — including the first version of this one — states it backwards. Telegram counts its limits in characters: the Bot API asks for “1-4096 characters after entities parsing” in a message and “0-1024 characters after entities parsing” in a caption, and the client configuration words the caption pair as a UTF-8 length. A styled letter is one character, so 1,024 bold letters fit in a caption exactly as 1,024 plain ones do.

What does double is a different number. Every character from the mathematical blocks sits above U+FFFF and fills two UTF-16 code units, and Telegram measures entity offsets and lengths — where a bold span starts and stops — in those units, explicitly “not code points”. That number matters if you are building messages through the API. It is not the limit, and a tool that warns you at 512 styled letters is enforcing a rule Telegram does not have. The counter above shows both; small caps, circled and fullwidth stay in the basic range and do not double even the second one.

Two things the generators quietly get wrong

Some letters have no styled twin, because Unicode had already encoded them elsewhere and refused to do it twice. Script B is not in the script block at all — it is ℬ, a letterlike symbol from 1991 — and the same is true of ℰ ℱ ℋ ℐ ℒ ℳ ℛ ℯ ℊ ℴ, of ℭ ℌ ℑ ℜ ℨ in fraktur, of ℂ ℍ ℕ ℙ ℚ ℝ ℤ in double-struck and of italic ℎ. A generator that just adds an offset emits an unassigned code point for those and you get an empty box. This one maps each of them explicitly.

And small caps has a genuine hole: there is no small capital X in Unicode, so an X stays an X. That is not a bug to work around — substituting a lookalike from another script is how text becomes unsearchable in a second, quieter way.

How to use them without looking like a spam channel

One accent, not a whole paragraph. Keep the searchable form of your name in the title, put the styled word beside it. Test the result on a second device before you commit it to a bio you will not revisit. And when the text is a post rather than a label, reach for real formatting instead.

---

Sources, checked September 2026: Telegram’s Bot API for the list of native entity types, for the 4,096-character message limit and the 1,024-character caption limit — both given as “characters after entities parsing” — and for entity offsets being counted in UTF-16 code units; message entities, which scopes that UTF-16 counting to “the length and offsets of entities”, explicitly “not code points”; Telegram’s client configuration for the 1,024/4,096 caption pair and the 70-character bio limit (140 with Premium), worded there as a maximum UTF-8 length, and spelled out as “UTF-8 characters” where the same page states a limit in full; and the Telegram FAQ for the username character set — “a-z, 0-9 and underscores” — which is why none of these letters can go in a username. Character mappings follow the Unicode Standard’s Mathematical Alphanumeric Symbols, Letterlike Symbols, Enclosed Alphanumerics, Halfwidth and Fullwidth Forms, IPA Extensions and Phonetic Extensions blocks.