Which AI model fools people most?

Live ranking from real quiz answers. Genuine photographs are fixed at 1500 — anything above that line gets picked as “the real one” more often than reality does.

Limited to the models built into this quiz — not every model on the market, and not always the latest version of each. A model's absence here says nothing about how good it is.

Image models

31 models listed · 993 AI images in the quiz · click a column to sort

			Confidence
—	Real image	1500	—	104811	65569	63%	—
1	Nano Banana 2 with Web SearchGoogle	1366	1351–1382	2567	625	24%	10
2	Ideogram 4.0 QualityIdeogram	1359	1344–1373	3234	750	23%	10
3	Nano Banana 2 LiteGoogle	1348	1335–1362	3455	791	23%	13
4	Imagen 2Google	1344	1335–1353	14609	2743	46%	29
5	Recraft V3Recraft	1338	1319–1357	1839	432	23%	13
6	Grok Imagine Image QualityxAI	1332	1313–1351	2029	417	21%	10
7	Qwen Image 2.0 ProAlibaba	1320	1305–1336	3020	603	20%	10
8	Reve 2.1Reve AI	1305	1290–1321	3204	601	19%	10
9	Juggernaut Pro FluxRunDiffusion	1273	1265–1282	13718	2221	19%	32
10	FLUX.2 [klein] 9BBlack Forest Labs	1268	1247–1289	2321	363	21%	14
11	Z-ImageAlibaba	1267	1262–1272	41594	6686	16%	132
12	Kling Image V3Kuaishou	1266	1249–1283	3061	488	16%	10
13	Seedream 4.5ByteDance	1240	1233–1247	17818	2544	14%	33
14	Stable Diffusion 3.5 LargeStability AI	1233	1218–1247	4593	645	14%	10
15	Stable DiffusionStability AI	1218	1208–1228	10902	1335	17%	37
16	Gemini 2.5 Flash Image / Nano BananaGoogle	1213	1202–1224	8394	1054	13%	14
17	FLUX.2 [pro]Black Forest Labs	1210	1188–1231	2435	284	12%	13
18	Imagen 3 FastGoogle	1208	1158–1259	414	52	13%	10
19	Imagen 4 UltraGoogle	1185	1163–1206	2775	283	10%	10
20	FLUX.2 [max]Black Forest Labs	1179	1164–1195	5295	548	10%	10
21	Kling IMAGE 3.0 OmniKuaishou	1177	1165–1188	9374	945	10%	24
22	SDXL-LightningByteDance	1171	1162–1179	19547	1791	9%	72
23	FLUX.1 [schnell]Black Forest Labs	1171	1166–1176	48849	4603	9%	157
24	GPT Image 1.5OpenAI	1169	1160–1179	13385	1290	10%	23
25	MAI-Image-2.5Microsoft AI	1156	1135–1178	3168	269	8%	10
26	Juggernaut Flux LightningRunDiffusion	1154	1135–1172	4009	370	9%	13
27	FLUX.1 [dev]Black Forest Labs	1152	1143–1161	18570	1606	9%	41
28	Seedream 4ByteDance	1148	1131–1166	4631	407	9%	13
29	DALL·E 3OpenAI	1143	1125–1161	4972	373	8%	27
30	Seedream 5.0 ProByteDance	1104	1042–1166	470	32	7%	10
31	Qwen-Image-MaxAlibaba	1084	1062–1107	3854	243	6%	12

Dimmed rows have fewer than 60 appearances — early evidence, not yet a settled rating.

A model joins this table once it has at least 10 images in the quiz and has appeared in at least 30 answers — below that, the number would say more about one lucky (or unlucky) image than about the model itself. New models and images are added regularly, so more entries — and firmer ratings — are on the way.

In the quiz but not yet listed: FLUX.2 LoRA Gallery: Realism (fal), FLUX1.1 [pro] (Black Forest Labs), GPT Image 2 Medium (OpenAI), Imagen 4 (Google), Lucid Realism (Leonardo AI), Titan Image Generator G1 v1 (Amazon) — fewer than 10 images in the quiz so far.

Last recomputed 14 Oct 2026, 03:15 · refreshed nightly

Add your own answers

Every round you play feeds straight into these numbers.

Architecture and geometry

Paintings

Brushwork and composition

Cars

Reflections and badges

🎬 Video quizzes

Not in the ranking above yet — too few answers so far to rate video models.

Gravity, fire, liquids

How the rating works

Bradley-Terry, not a win count. Each round counts as a choice among the options actually on screen, so beating strong rivals is worth more than beating weak ones. Two-option rounds and four-option rounds therefore combine without their different odds distorting anything.

Read the range, not the position. Where two intervals overlap, the order between those models is not established. Dimmed rows have fewer than 20 appearances and are listed for completeness only.

Pool size is context. A model represented by three images is judged on those three images, however often they were shown.

Replays don't get extra weight — within a visit. Only your first look at a round counts, because after that you'd recognise it and it stops being a fresh judgment. That rule ends when you close the tab: come back later the same day and a repeat is counted again.

Limits. The method assumes all players judge alike, and this measures our image pool rather than a model's ceiling. Answers recorded before 29 August 2026 come only from players who accepted analytics cookies; since then every player counts, so figures spanning that date mix two different groups.