Metraipolis › archive › dc_d11fe75cdfaa
Distinguishing among reasoning levels of GPT 5.5 Pro via a grammar lens benchmark — arXiv:2611.00555
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.00555
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-03T18:22:37.583Z
- SHA-256 of the archived text
d9ea424eaa02201404cfcdda113f3483a228c165f866bb2260c094a018dccff4- Stable URL of this page
https://metraipolis.com/documento/dc_d11fe75cdfaa
How to cite
Distinguishing among reasoning levels of GPT 5.5 Pro via a grammar lens benchmark — arXiv:2611.00555 — arXiv (en), https://arxiv.org/abs/2611.00555 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_d11fe75cdfaa (Metraipolis archive dc_d11fe75cdfaa, sha256 d9ea424eaa02201404cfcdda113f3483a228c165f866bb2260c094a018dccff4)
BibTeX
@misc{dcd11fe75cdfaa,
title = {Distinguishing among reasoning levels of GPT 5.5 Pro via a grammar lens benchmark — arXiv:2611.00555},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.00555},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_d11fe75cdfaa; sha256 d9ea424eaa02201404cfcdda113f3483a228c165f866bb2260c094a018dccff4},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.00555. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Does the level of test-time processing effort affect a language model's computational capability? A formal grammar-lens benchmark method, derived from the theory of computation, generates novel data at specified levels, each successively more computationally powerful than the levels below. Using this data, we tested GPT-5.5, and GPT-5.5-Pro-Medium, -High, and -XHigh, showing incrementally increased acceptance of test data: 16, 40, 43, and 52\%, respectively, illustrating the benchmark's ability to discern fine-grained differences among models' added test-time reasoning levels. Additionally, as test-time compute effort increases, model responses diverge from human responses to the same data. This increasing disagreement may turn out to be broadly characteristic of more capable language models, introducing the potential of a general capability-versus-alignment conflict. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno). Categorie arXiv: Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2611.00555 — 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 2026-11-03T18:10:07.889Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).