Skip to content

Image models for fantasy and tabletop work

Last vote 20 Nov 2026684 counted blind votes17 ranked · 73 in evaluationMethodologyDownload data (CSV)How to cite

73 of 90 models still need votes to be ranked.

Checking your account…

Standings · Overall

Pick a model to see each task. Sorted by score · ranges are 95%
Overall ranking of 17 models. Scores show a 95% range.
RankModelScore95% rangeVotesWon–lost–tiedWin rate
11–51,239 ±70147108–31–876%
21–8Preliminary1,207 ±1162015–4–178%
31–81,166 ±7112782–39–667%
41–91,125 ±6312677–39–1065%
51–13Preliminary1,102 ±913621–11–464%
62–131,075 ±679651–38–757%
72–131,056 ±945231–21–060%
82–14Preliminary1,041 ±963215–15–250%
94–16Preliminary1,002 ±833114–15–248%
105–17Preliminary967 ±863412–18–441%
115–17963 ±709839–57–241%
125–17958 ±846928–39–242%
135–17915 ±1045017–33–034%
148–17Preliminary887 ±794310–29–428%
159–17853 ±995714–42–125%
169–17Preliminary841 ±102398–29–223%
1710–17833 ±778116–59–623%

How the ranking works

Signed-in CharGen users compare two outputs for one fantasy prompt with the model names hidden. Scores are a Bradley-Terry fit with 95% ranges; skips and “both poor” do not count. Full methodology

Monthly snapshot

CharGen Fantasy Arena, October 2026: a frozen, dated copy of every board with ranges and vote counts, CC BY 4.0. CSV · JSON

How to cite

CharGen Fantasy Arena, Image board, October 2026 snapshot. char-gen.com/ai-model-leaderboard/image