arXiv cs.AI· Minghao Li, Bangyan Li, Zifan Wang, Yulong Li, Hu Xu, Gan Zhang, Jingtong Wu, Wenqiang Xu·· 9 天前AI 评分32
FlowState：以执行状态作为长程 LLM 智能体的记忆
FlowState: Execution State as Memory for Long-Horizon LLM Agents
AI 导读
FlowState 将执行状态作为可跨请求保留与重访的记忆，通过增量状态更新（ISU）与渐进状态访问（PSA）统一当前决策与历史信息复用。在相同 DeepSeek-V4-Flash 模型下，其 MemoryArena 平均成功率与 τ³-Bench 平均通过率分别比全上下文基线提升 4.55 和 13.95 个百分点，总 token 消耗降低 43.2% 和 40.6%。
正文 · 原文
Abstract:Long-horizon tasks require LLM agents to continually draw on information from earlier interactions. However, retaining the full history increases context costs, while compressing it risks losing details needed later, and the relevance of historical information often becomes apparent as the task progresses. To address these challenges, we propose FlowState, which treats execution state as memory that can be retained and revisited across requests, unifying current decision-making with the reuse of historical information. FlowState preserves semantically typed state nodes, their relations, and references to raw tool observations, separating persistent retention from on-demand access. Within a single execution loop, Incremental State Update (ISU) maintains the current state based on new inputs and feedback, while Progressive State Access (PSA) progressively reveals historical states and supporting evidence as needed during reasoning. Together, these mechanisms enable agents to reassess prior decisions in light of new information and guide subsequent actions. Compared with a full-context baseline using the same DeepSeek-V4-Flash model, FlowState improves the average success rate on MemoryArena and the average pass rate on $\tau^3$-Bench by 4.55 and 13.95 percentage points, respectively, while reducing total token consumption by 43.2% and 40.6%. These results demonstrate the performance and efficiency advantages of FlowState on long-horizon tasks.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2609.34565 [cs.AI] |
| (or arXiv:2609.34565v2 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2609.34565 arXiv-issued DOI via DataCite |
Submission history
From: Bangyan Li [view email]
[v1]
Mon, 28 Sep 2026 08:18:03 UTC (328 KB)
[v2]
Thu, 8 Oct 2026 12:31:23 UTC (328 KB)
来源：arXiv cs.AI · arxiv.org