Metraipolis › archive › dc_e4fa4950c13d

Beyond Token Savings: A Systematic Study of Context Compression in LLM Agents — arXiv:2609.32961

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.32961
License (as archived)
CC0-1.0
Archived
2026-10-07T21:00:51.753Z
SHA-256 of the archived text
822c9e03deab5210fe7d0165bdf411815318afe4274993e0f297e21ed1321163
Stable URL of this page
https://metraipolis.com/documento/dc_e4fa4950c13d

How to cite

Beyond Token Savings: A Systematic Study of Context Compression in LLM Agents — arXiv:2609.32961 — arXiv (en), https://arxiv.org/abs/2609.32961 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e4fa4950c13d (Metraipolis archive dc_e4fa4950c13d, sha256 822c9e03deab5210fe7d0165bdf411815318afe4274993e0f297e21ed1321163)
BibTeX
@misc{dce4fa4950c13d,
  title = {Beyond Token Savings: A Systematic Study of Context Compression in LLM Agents — arXiv:2609.32961},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.32961},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e4fa4950c13d; sha256 822c9e03deab5210fe7d0165bdf411815318afe4274993e0f297e21ed1321163},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.32961. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

As LLM agents tackle longer tasks, they increasingly compress growing histories of reasoning, actions, and tool outputs. Compression can reduce token use, but it also changes the information available for later decisions. Existing agentic harnesses bundle decisions about what to compress, when to compress, and how much to remove into fixed policies. A systematic characterization is needed to disentangle these decisions and reveal how each affects task success and execution cost. We systematically vary these decisions across three open-weight models on SWE-bench Verified and Terminal-Bench 1.0. Across nearly 35,000 agent runs, we measure task success, token use, end-to-end latency, and estimated cost. We find that fewer tokens need not mean faster or cheaper execution: on Terminal-Bench with Qwen, policies using roughly one-third as many tokens can take 20-80% longer than the uncompressed agent. Policies with similar overall success can solve different tasks, while the same policy can perform quite differently across models. Our results motivate evaluating compression by its effects on agent execution and tailoring policies to the task, model, and workload.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.AI).
Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.32961 — 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 2026-10-07T20:29:02.515Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).