ObsDriveBench - Observability Awareness: leaderboard
Judge what each sensor can still observe under fog, rain, snow or darkness and whether camera, LiDAR and radar agree; from ObsDriveBench (2026), built on K-Radar. Accuracy in percent, higher is better.
Metric: Accuracy (%; dimension average of modality-specific observability, cross-modal comparison and cross-modal verification; multiple-choice questions on synchronized camera, LiDAR BEV and radar BEV inputs from K-Radar under fog, rain, snow and low light, answers matched by an LLM judge; official inference settings). Source: arxiv.org. Saturation forecast: Around 2030. 15 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 54.72 | #94 |
| 2 | InternVL3-8B | 44.8 | #1688 |
| 3 | Qwen 2.5 VL 7B | 43.74 | #1787 |
| 4 | GPT-5 | 41.89 | #167 |
| 5 | InternVL3-38B | 40.82 | #903 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=obsdrivebench-observability-awareness · How It Works · Data refreshed daily, snapshot 2026-11-13.