360CityArena - Relational Spatial Reasoning: leaderboard
360CityArena (2026) places a multimodal agent in a navigable reconstruction of the Akihabara district of Tokyo built from 360-degree street video, where it explores with discrete view and movement actions while a map marks its position. In this task the agent names the landmark that stands in a given spatial relation to a reference landmark, judged by GPT-5; accuracy over 25 tasks, in percent; higher is better.
Metric: Accuracy (%; the agent names the landmark that stands in a given spatial relation to a reference landmark, judged semantically by GPT-5 and checked by hand; 25 human-authored tasks in a navigable reconstruction of the Akihabara district from 602 360-degree video segments; the agent acts with seven discrete view and movement actions and sees a map marking its position and heading; one run). Source: arxiv.org. Saturation forecast: Around February 2027. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 32 | #167 |
| 2 | Gemini 2.5 Flash | 12 | #453 |
| 3 | Claude Sonnet 4.5 | 8 | #250 |
| 4 | Qwen 2.5 VL 32B Instruct | 4 | #1053 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=360cityarena-relational-spatial-reasoning · How It Works · Data refreshed daily, snapshot 2026-11-13.