Metraipolis › archive › dc_bb369f6134ae
StateSight: Benchmarking Latent Spatial-State Reconstruction in Vision-Language Models — arXiv:2608.20414
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.20414
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-18T10:56:27.424Z
- SHA-256 of the archived text
853a8b358a0ac9f30e8bffc7e7e6e7bea935f4cc353dd0e6862f05b6d705519c- Stable URL of this page
https://metraipolis.com/documento/dc_bb369f6134ae
How to cite
StateSight: Benchmarking Latent Spatial-State Reconstruction in Vision-Language Models — arXiv:2608.20414 — arXiv (en), https://arxiv.org/abs/2608.20414 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_bb369f6134ae (Metraipolis archive dc_bb369f6134ae, sha256 853a8b358a0ac9f30e8bffc7e7e6e7bea935f4cc353dd0e6862f05b6d705519c)
BibTeX
@misc{dcbb369f6134ae,
title = {StateSight: Benchmarking Latent Spatial-State Reconstruction in Vision-Language Models — arXiv:2608.20414},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.20414},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_bb369f6134ae; sha256 853a8b358a0ac9f30e8bffc7e7e6e7bea935f4cc353dd0e6862f05b6d705519c},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.20414. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Vision-language models are increasingly used for multimodal question answering, yet their ability to reconstruct latent spatial structure from a single image remains difficult to isolate. Broad benchmarks often combine perception, optical character recognition, domain knowledge, linguistic priors, and reasoning in the same evaluation. We introduce StateSight, a procedurally generated benchmark for cube-net opposite-face reasoning, occluded cube-tower counting, and 4-neighbor connected-component counting. Each task family contains 300 single-image prompts with deterministic oracle labels and exact-match scoring. OpenAI GPT-5.5, using the API model identifier gpt-5.5, achieved 59.3%, 33.3%, and 28.3% accuracy across the three tasks, while Claude Sonnet 5 achieved 53.3%, 18.7%, and 7.3%. All final direct runs had zero format errors. A 30-participant human baseline on 60 items exceeded both models on every task, with mean accuracies of 80.8%, 68.8%, and 64.3%. Visible-derivation analysis identified recurring errors in image-state reconstruction and reasoning procedure. We also introduce StateSight-Steps, a companion dataset of 900 interleaved image-text examples and 3,600 deterministic intermediate visual states. The results show that format-valid responses can mask failures to recover the spatial structure required for verifiable visual inference. Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-08-24 in cs.CV). Categorie arXiv: Artificial Intelligence (cs.AI) ; Computer Vision and Pattern Recognition (cs.CV). Pagina del lavoro: https://arxiv.org/abs/2608.20414 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-18T10:48:13.614Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-08-24?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).