VGenST-Bench - Height Ordering: leaderboard

VGenST-Bench (2026) tests spatio-temporal reasoning of multimodal LLMs on short videos synthesized from validated scene graphs with text-to-image and image-to-video models. Height Ordering task: order the objects in a room by height (vista scale, exocentric, static). Accuracy in percent on multiple-choice questions under circular evaluation (right under every rotation of the options), 8 sampled frames; higher is better.

Metric: Height Ordering (vista scale, exocentric, static) accuracy (%) on the base multiple-choice questions of VGenST-Bench under circular evaluation (correct only if right under every cyclic permutation of the options), macro-averaged over the task's QA types, 8 uniformly sampled frames from about 100 synthesized videos per task (generated with text-to-image and image-to-video models from validated scene graphs); higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 15 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemini 3 Flash96.9#180
2Gemini 3.1 Flash Lite87.1#377
3Kimi K2.686.1#197
4GPT-5.484.5#138
5GLM-4.6V-Flash78.6#1049
6GPT-5.4 Mini70.7#409
7Qwen 3.5 27B69.5#380
8Qwen 3.5 9B67.2#838
9Gemma 4 31B (IT)66.8#354
10InternVL3.5-30B-A3B64.7#1269
11Qwen 3.5 4B63.6#1270
12Gemma 4 26B A4B (IT)61.9#563
13InternVL3.5-8B51.1#1618
14InternVL3.5-4B47.6#1946
15GPT-5.4 Nano44#653

No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=vgenst-bench-height-ordering · How It Works · Data refreshed daily, snapshot 2026-11-22.