PinpointQA - Nearest Reference Identification: leaderboard
Metric: Exact-match accuracy (times 100, micro over QA pairs) of four-option nearest reference object identification (excluding the supporting surface) of the PinpointQA evaluation split (2,019 QA pairs from held-out ScanNet++ and ScanNet200 scenes; 64 uniformly sampled RGB frames per indoor video, no depth or 3D input, greedy decoding); higher is better. Source: arxiv.org. Saturation forecast: Around 2029. 8 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Kimi K2.5 | 42 |
| 2 | GPT-5.4 | 39 |
| 3 | Qwen 3 VL 8B Instruct | 37 |
Interactive version: theaggregate.ai/benchmark?slug=pinpointqa-nearest-reference-identification · How It Works · Data refreshed daily, snapshot 2026-10-07.