Solo practice
Sixty per cent of every game ever logged is one person practising alone, and the leaderboard, the win rates and the head-to-head records all throw it away on purpose — there is no opponent, so none of those numbers mean anything here. This page is the other half: 35 games and 1,124 turns that only measure a player against the dictionary.
Wins, losses, placement and win rate are absent below rather than shown as zero. A solo game has one participant, so those metrics have no denominator at all.
Volume over time
Solo turns per day
View as table
| Month | Turns | Games |
|---|---|---|
| March 2026 | 11,595 | 253 |
| April 2026 | 12,292 | 318 |
| May 2026 | 4,199 | 65 |
| June 2026 | 2,106 | 29 |
| July 2026 | 3,096 | 72 |
| August 2026 | 6,263 | 109 |
| September 2026 | 309 | 9 |
Who practises
Ranked on turns played — there is no rate here worth ranking on
| # | Player | Turns | ||
|---|---|---|---|---|
| 1st | 401 | 99.0%n=401 | ixsbrutal#7,198 | |
| 2nd | 296 | 86.8%n=296 | bpirare#6,344 | |
| 3rd | 169 | 89.9%n=169 | voprare#5,940 | |
| 4th | 130 | 78.5%n=130 | nzyuncommon#3,695 | |
| 5th | 128 | 97.7%n=128 | layfamiliar#1,390 |
“Survived” is the share of turns that ended in an accepted word. tea-bot never logs a rejected guess, so it cannot measure first-guess correctness. “Hardest prompt” is that player’s highest-ranked prompt by position in the bot’s own prompt list, which is ordered easiest to hardest — an objective difficulty, not one inferred from a handful of attempts.
Longest sittings
Consecutive practice with no break longer than 30m
- 1378turns29:11unbroken2games240best run99.5%survived17 Mar 2026 · 6 months ago
- 2189turns23:10unbroken9games84best run83.6%survived17 Mar 2026 · 6 months ago
- 3126turns8:46unbroken1games71best run98.4%survived17 Mar 2026 · 6 months ago
- 4107turns10:50unbroken2games41best run92.5%survived17 Mar 2026 · 6 months ago
- 582turns40:05unbroken2games20best run73.2%survived17 Mar 2026 · 6 months ago
- 677turns10:25unbroken5games26best run87.0%survived17 Mar 2026 · 6 months ago
- 765turns5:57unbroken1games48best run96.9%survived17 Mar 2026 · 6 months ago
- 826turns4:05unbroken2games14best run84.6%survived17 Mar 2026 · 6 months ago
- 925turns2:27unbroken1games12best run92.0%survived17 Mar 2026 · 6 months ago
- 1023turns3:11unbroken2games11best run82.6%survived17 Mar 2026 · 6 months ago
- 1123turns2:18unbroken2games19best run91.3%survived17 Mar 2026 · 6 months ago
- 122turns0:15unbroken1games1best run50.0%survived17 Mar 2026 · 6 months ago
A sitting is not a game — practice runs straight from one game into the next, and the longest here covers 2 games. The 30m gap is a judgement call: a different threshold would draw different boundaries, so it is stated rather than assumed.
What difficulty gets practised
Prompts by their position in the bot's own easiest-to-hardest list
| Tier | Turns | Share of solo turns | Survived |
|---|---|---|---|
| common | 238 | 21.2% | 96.2% |
| familiar | 446 | 39.7% | 94.8% |
| uncommon | 310 | 27.6% | 90.6% |
| rare | 126 | 11.2% | 77.8% |
| brutal | 3 | 0.3% | 66.7% |
Difficulty is the prompt’s line number in the bot’s prompt list, which is ordered easiest first, so “brutal” means rank 7,001 and above out of 9,217. 1 solo turns (0.1%) used a prompt that is not in that list — mostly multi-word and generated prompts — and are excluded from this table rather than filed under a tier they do not have.
Hardest prompts solved alone
Ranked by objective difficulty, with the attempt count beside every one
| Prompt | Difficulty | First solved by | ||
|---|---|---|---|---|
| ixs | #7,198 / 9,217 | brutal | SIXSCORE | |
| xyf | #7,178 / 9,217 | brutal | OXYFLUORIDE | |
| bpi | #6,344 / 9,217 | rare | BPI | |
| wsa | #6,189 / 9,217 | rare | BROWSABILITY | |
| ltb | #6,109 / 9,217 | rare | SALTBUSH | |
| oxs | #6,098 / 9,217 | rare | FOXSKIN | |
| tzk | #6,066 / 9,217 | rare | BRITZKA | |
| ykn | #6,064 / 9,217 | rare | GRANNYKNOT | |
| paa | #6,063 / 9,217 | rare | PAAL | |
| tzp | #6,004 / 9,217 | rare | CHUTZPAH | |
| fbe | #5,967 / 9,217 | rare | OFFBEAT | |
| ysg | #5,964 / 9,217 | rare | DYSGENIC | |
| emc | #5,948 / 9,217 | rare | EMCEE | |
| vop | #5,940 / 9,217 | rare | OVOPLASM | |
| mdr | #5,922 / 9,217 | rare | GUMDROP | |
| eok | #5,913 / 9,217 | rare | PALEOKINETIC | |
| eax | #5,877 / 9,217 | rare | SEAX | |
| toj | #5,807 / 9,217 | rare | PHOTOJOURNALISM | |
| ylh | #5,806 / 9,217 | rare | PHENYLHYDRAZINE | |
| uzo | #5,793 / 9,217 | rare | BOUZOUKI | |
| uby | #5,778 / 9,217 | rare | RUBY | |
| yus | #5,765 / 9,217 | rare | YUS | |
| mbh | #5,738 / 9,217 | rare | DUMBHEAD | |
| wnd | #5,704 / 9,217 | rare | DOWNDRAFT |
Nearly all of these were seen exactly once, which is why the attempt count sits in the table rather than a solve rate: “1 of 1” is honest, and “100%” would dress one lucky answer up as mastery. Prompts past #9,165 are no longer offered by the bot at all.
Everything on this page comes from the same replica of the bot’s logs as the rest of the site, restricted to games with exactly one participant. Average answer time across all solo practice is 3.9 s, and the average practice game runs 2:47 over 36.3 turns. 4 of those games logged no turns at all: red, green and yellow tea score whole rounds rather than individual turns, so the bot writes round submissions for them and there is nothing here to count. They are included in the game total and excluded from every turn-based figure, rather than quietly dropped from both.