SpatialGen-Bench (Text Answering): leaderboard
Spatial cognition benchmark from the ProVisE project (ZJU-OmniAI, "Show, Don't Tell"): perception, understanding, reasoning, and interaction tasks answered in text; the unweighted mean of all 14 integrated task accuracies, combining audited core results with the three expansion tasks. Only source-complete runs receive an overall.
Metric: Overall Accuracy (%). Source: zju-omniai.github.io. Saturation forecast: Around February 2027. 27 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.8 Max | 81.64 | #34 |
| 2 | Claude Fable 5 | 81.2 | #20 |
| 3 | GPT-5.6 Sol | 80.98 | #22 |
| 4 | Kimi K3 | 63.22 | #35 |
| 5 | GPT-5.4 | 60.87 | #83 |
| 6 | Claude Opus 4.8 | 57.64 | #39 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=spatialgen-bench-text-answering · How It Works · Data refreshed daily, snapshot 2026-11-05.