Metraipolis › archive › dc_e373f98fa448
Kepler: Auditable World Models for ARC-AGI-3 — arXiv:2610.00834
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.00834
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-06T11:54:41.468Z
- SHA-256 of the archived text
2e4eda86b6dbe6851c0a86addff92159eb0810039e5bbc0628b0c851468e64d0- Stable URL of this page
https://metraipolis.com/documento/dc_e373f98fa448
How to cite
Kepler: Auditable World Models for ARC-AGI-3 — arXiv:2610.00834 — arXiv (en), https://arxiv.org/abs/2610.00834 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e373f98fa448 (Metraipolis archive dc_e373f98fa448, sha256 2e4eda86b6dbe6851c0a86addff92159eb0810039e5bbc0628b0c851468e64d0)
BibTeX
@misc{dce373f98fa448,
title = {Kepler: Auditable World Models for ARC-AGI-3 — arXiv:2610.00834},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.00834},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e373f98fa448; sha256 2e4eda86b6dbe6851c0a86addff92159eb0810039e5bbc0628b0c851468e64d0},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.00834. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
ARC-AGI-3 evaluates agents in interactive environments whose rules and objectives must be inferred from observation. We present Kepler, an open-source harness that represents hypotheses as executable world models and validates them through retrospective transition checks and conditional prediction checks. Under one frozen Claude Opus 5 configuration, Kepler obtained a server-verified 100.00 RHAE on all 25 public games, with no per-game model selection or score-conditioned reruns. On 181 of 183 completed levels, the final Opus attempt used no more actions than the corresponding median-human baseline. The retained board runs used 8,256 environment actions, of which 7,292 occurred in scored levels. Retained local provider-session records yield 858.0 million tokens, 97.37% cache reads, and a \$777.72 cost at September 1, 2026 API list-equivalent rates. We also report three evaluation failures: source-code leakage that produced an invalid perfect run, agents reconstructing a removed harness in a control condition, and autonomous repair masking a broken planner. A single-game observation case study showed that animation frames contained task-relevant information absent from settled text grids. Across the final Claude Opus 5 and GPT-5.6 Sol boards, 48 of 50 game-model cells reached 100. These results indicate that public-set score alone has limited discriminative value and motivate first-attempt, cost-conditioned, and verification-aware reporting. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-10-02 in cs.MA). Categorie arXiv: Artificial Intelligence (cs.AI) ; Multiagent Systems (cs.MA). Pagina del lavoro: https://arxiv.org/abs/2610.00834 — 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 2026-10-06T10:17:30.562Z dalla pagina https://arxiv.org/catchup/cs.MA/2026-10-02?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).