Metraipolis › archive › dc_cc2acfb76e8b

Staying on Task: Testing the Foundations of Long-Horizon Agent Reliability — arXiv:2609.38712

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.38712
License (as archived)
CC0-1.0
Archived
2026-10-07T12:42:05.305Z
SHA-256 of the archived text
1ec78120423b5e512f0e266a2354e69600d327a43259b96beb4a9062a9352cc6
Stable URL of this page
https://metraipolis.com/documento/dc_cc2acfb76e8b

How to cite

Staying on Task: Testing the Foundations of Long-Horizon Agent Reliability — arXiv:2609.38712 — arXiv (en), https://arxiv.org/abs/2609.38712 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_cc2acfb76e8b (Metraipolis archive dc_cc2acfb76e8b, sha256 1ec78120423b5e512f0e266a2354e69600d327a43259b96beb4a9062a9352cc6)
BibTeX
@misc{dccc2acfb76e8b,
  title = {Staying on Task: Testing the Foundations of Long-Horizon Agent Reliability — arXiv:2609.38712},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.38712},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_cc2acfb76e8b; sha256 1ec78120423b5e512f0e266a2354e69600d327a43259b96beb4a9062a9352cc6},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.38712. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Long-horizon agentic workflows require models to sustain repeated state-dependent actions all while the context grows, sub-task complexity changes, and new data arrives. Each situation represents an independent axis along which an agent may fail. An agent reconciling a long ledger, for example, must repeatedly read its state, update the correct record, and preserve alignment across thousands of outputs. A model may accept the entire ledger yet lose its place or stop applying the operation consistently as generation proceeds. We introduce Long-Transduction, a controlled diagnostic that tests a model's ability to stay on task during long generation while continuously reading, mutating, and outputting input-context dependent operations such as arithmetic, sorting, variable lookups, and table transformations. Long-Transduction evaluation independently varies local task complexity, input data formatting, and context length isolate failures along each axis. We evaluate seven open-weight models, finding a 62.8\% decrease when scaling context length from 4-128K, a 36.5\% decrease when varying input format, and a 39.9\% decrease by increasing local task complexity. Together, these failures represent critical liabilities in long-horizon agentic workflows.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-10-01 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.38712 — 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 2026-10-07T12:14:13.668Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-10-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).