Parallel Context Compaction for Long-Horizon LLM Agent Serving

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Cim, Musa, Topcu, Burak, Das, Chita, Kandemir, Mahmut Taylan
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866914591100895232
author Cim, Musa
Topcu, Burak
Das, Chita
Kandemir, Mahmut Taylan
author_facet Cim, Musa
Topcu, Burak
Das, Chita
Kandemir, Mahmut Taylan
contents Long-horizon LLM agents accumulate growing conversation histories that eventually exceed the model's context window. Context compaction via LLM-based summarization keeps the conversation bounded, but summarization is inherently lossy and the blocking call stalls agent inference for tens of seconds. Moreover, the operator has no fine-grained control over summary volume since prompt instructions are largely ignored, and as context grows, both the amount of output tokens the model produces and the information it retains fluctuate substantially from run to run, making the agent's retained knowledge unpredictable across runs. We introduce \textbf{parallel compaction} for long-horizon agentic flows and characterize it against the sequential synchronous baseline across four backbones spanning 8B to 120B parameters, mixing dense and MoE architectures with reasoning and non-reasoning models, on the HotpotQA multi-hop QA and LoCoMo long-context dialogue benchmarks. Parallel compaction gives the operator fine-grained, predictable control over summary volume and enables more targeted prompt engineering per block. At matched compaction decode volume, it reduces end-to-end wall time and improves compaction throughput over the sequential baseline.
format Preprint
id arxiv_https___arxiv_org_abs_2605_23296
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Parallel Context Compaction for Long-Horizon LLM Agent Serving
Cim, Musa
Topcu, Burak
Das, Chita
Kandemir, Mahmut Taylan
Artificial Intelligence
Long-horizon LLM agents accumulate growing conversation histories that eventually exceed the model's context window. Context compaction via LLM-based summarization keeps the conversation bounded, but summarization is inherently lossy and the blocking call stalls agent inference for tens of seconds. Moreover, the operator has no fine-grained control over summary volume since prompt instructions are largely ignored, and as context grows, both the amount of output tokens the model produces and the information it retains fluctuate substantially from run to run, making the agent's retained knowledge unpredictable across runs. We introduce \textbf{parallel compaction} for long-horizon agentic flows and characterize it against the sequential synchronous baseline across four backbones spanning 8B to 120B parameters, mixing dense and MoE architectures with reasoning and non-reasoning models, on the HotpotQA multi-hop QA and LoCoMo long-context dialogue benchmarks. Parallel compaction gives the operator fine-grained, predictable control over summary volume and enables more targeted prompt engineering per block. At matched compaction decode volume, it reduces end-to-end wall time and improves compaction throughput over the sequential baseline.
title Parallel Context Compaction for Long-Horizon LLM Agent Serving
topic Artificial Intelligence
url https://arxiv.org/abs/2605.23296