VGenST-Bench - Directional Signage Grounding: 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. Directional Signage Grounding task: use directional signs to tell which way leads to a target (environmental scale, egocentric, 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: Directional Signage Grounding (environmental scale, egocentric, 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
1GPT-5.497#138
2Gemini 3 Flash97#180
3Qwen 3.5 27B91.9#380
4Gemini 3.1 Flash Lite91#377
5Qwen 3.5 9B90.9#838
6Qwen 3.5 4B88#1270
7Gemma 4 31B (IT)86.6#354
8Gemma 4 26B A4B (IT)81.6#563
9Kimi K2.679.8#197
10InternVL3.5-30B-A3B77.8#1269
11GPT-5.4 Mini75#409
12GLM-4.6V-Flash70.5#1049
13InternVL3.5-8B64#1618
14InternVL3.5-4B60.5#1946
15GPT-5.4 Nano59.5#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-directional-signage-grounding · How It Works · Data refreshed daily, snapshot 2026-11-22.