Telegram username prices: what the market actually pays
Market context: TGGI 108.87 ▲0.01% · 2026-11-03 · the index
On 2026-11-02 the seven-day median sale price of a four-letter Telegram username on Fragment was 6,049 GRAM across 36 sales. Five- and six-letter handles both had a median of 10 GRAM (434 and 472 sales), and English dictionary words a median of 840 GRAM across 19 sales. Those four numbers are the market in one line: length sets the ceiling, and meaning decides who reaches it.
| Class | Sales in 7 days | Median, GRAM | What it tells you |
|---|---|---|---|
| 4 letters | 36 | 6,049 | The smallest supply. Brandable names are priced as brands. |
| 5 letters | 434 | 10 | Mostly random mixes sold in bulk near the market floor. |
| 6 letters | 472 | 10 | Same as five letters: high volume, cheap middle. |
| English words, any length | 19 | 840 | Meaning is scarce. Few sales, wide spread. |
Seven-day medians of recorded Fragment sales, 2026-11-02, from the site's own open file /data/usernames-daily.json (CC BY 4.0). A median of 10 GRAM does not mean every five-letter name is worth 10: it means most recorded sales in that class were cheap handles.
What actually drives the price
Length
Supply, not value. Four-letter handles are the scarcest class, so the median is set by the few brandable ones; a random four-letter mix still trades near the floor for random handles, which our method puts at 5–10 GRAM.
A dictionary word
Meaning is scarce, so it beats length: the median for English words was 840 GRAM, above five- and six-letter mixes. The catch is the sample — only 19 sales, so the spread is wide and one sale moves it.
A pattern
Palindromes, repeats and short numbers (@llllyl, @fgffg) are bought by collectors of the pattern itself. There are few comparable sales, so we price them from their own history only and never from the length class.
Brandable
A pronounceable short name, a two-word brand or a crypto keyword is worth what the cheapest similar handle is listed at today, not what its letters suggest. That competing listing is the ceiling, and it moves.
The 5% fee, and why the headline price is not what you keep
Fragment takes 5% of the sale price, so a seller receives 95%. A 1,000 GRAM sale returns 950 GRAM before network and conversion fees, which no figure on this site includes. Never compare a gross sale price with a net cost: that is how a flat trade starts to look like a profit.
How we value a username
Every handle is sorted into a class first — four characters, an English word, a palindrome or repeating pattern, a short number, a two-word brand, a crypto keyword, a pronounceable brand — and random letter mixes are marked as junk and never become a deal. Inside the class the value is a recency-weighted median of real Fragment sales over the last 12 months, with at least 8 sales after cleaning; a sale's weight halves every 60 days. The same handle changing hands again within 60 days is not a new datapoint, prices more than 8 times away from the class median are dropped, and brandable or pattern classes are capped by the cheapest similar handle listed right now. When the class has too few sales, the answer is “not enough comparable sales” with the last sale and its date — never a number invented to fill the space.
Questions
Is every four-letter username worth 6,049 GRAM?
No. 6,049 GRAM is the median across 36 sales of that class, pulled up by brandable names. A random four-letter mix trades near the floor for random handles, which the method puts at 5–10 GRAM.
Why is the median for five-letter names only 10 GRAM?
Because most recorded sales in that class are bulk cheap handles. A median describes the middle of a class, not its best members — which is why the estimate also cleans and caps the data before it answers.
Does the price include Fragment's 5% fee?
No, every number on this page is a gross sale price. Fragment takes 5% when you resell, so a 1,000 GRAM sale returns 950 GRAM before network and conversion fees.
How many sales do you need before a median means anything?
At least 8 after cleaning. Below that the page says “not enough comparable sales” and shows the last recorded sale with its date, because a median from two sales is an anecdote with a decimal point.
Medians: 2026-11-02 · Method and page: 2026-11-03