Metraipolis › archive › dc_fe33c77ef704

Think or Remember: The Exchange Rate of Agentic Inference — arXiv:2611.10982

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.10982
License (as archived)
CC0-1.0
Archived
2026-11-10T18:42:59.168Z
SHA-256 of the archived text
93d002522ecfc10012af3778a692c1fbd58ba26828a2ea937be007343ab666de
Stable URL of this page
https://metraipolis.com/documento/dc_fe33c77ef704

How to cite

Think or Remember: The Exchange Rate of Agentic Inference — arXiv:2611.10982 — arXiv (en), https://arxiv.org/abs/2611.10982 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_fe33c77ef704 (Metraipolis archive dc_fe33c77ef704, sha256 93d002522ecfc10012af3778a692c1fbd58ba26828a2ea937be007343ab666de)
BibTeX
@misc{dcfe33c77ef704,
  title = {Think or Remember: The Exchange Rate of Agentic Inference — arXiv:2611.10982},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.10982},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_fe33c77ef704; sha256 93d002522ecfc10012af3778a692c1fbd58ba26828a2ea937be007343ab666de},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.10982. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Agents built on reasoning models resend their growing transcript at every call. Two transacting parties, not the model, decide what that costs: the client keeps or edits messages; the server decides whether earlier reasoning returns in the prompt and what its cache reuses. We define a cost model whose counts match an instrumented engine exactly; the architecture charges $c$ per token processed and $\beta$ per pair attended. Their ratio $P^*=c/\beta$ is an exchange rate between thinking and remembering: per token, processing dominates below this context length, a thinking regime, and attention above it, a memory regime. A policy change has a mechanical cost, incurred even if the agent writes the same tokens, and a behavioral one: it writes different tokens, changing cost and outcome. Serving alone can flip which policy is cheaper: on recorded Qwen3-8B ScienceWorld runs, dropping earlier reasoning costs 1.74 times keeping it when the server reuses generated states and 0.34 times when it reuses nothing. In live reruns, dropping it cuts correct HotpotQA answers by 9-23 percentage points on four models, at lower point-estimate cost. Comparing context policies means comparing systems: architecture and serving set the cost of the history kept and the work it induces.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2611.10982 — 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 2026-11-10T18:37:37.618Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).