360CityArena - Object Count: leaderboard
Metric: Mean relative accuracy (%; count of specified objects in a designated area, credited over relative-error thresholds 0.50 to 0.95; 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 2036. 6 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Flash | 24 |
| 2 | Qwen 2.5 VL 32B Instruct | 18.8 |
| 3 | Claude Sonnet 4.5 | 10.8 |
| 4 | InternVL3.5-8B | 2.8 |
| 5 | GPT-5 | 2.4 |
Interactive version: theaggregate.ai/benchmark?slug=360cityarena-object-count · How It Works · Data refreshed daily, snapshot 2026-09-29.