AOS-Bench：LLM Agent 能否管理动态工作负载？
AOS-Bench: Can LLM Agents Manage Dynamic Workloads?
AI 导读
AOS-Bench 基准将 LLM Agent 系统作为工作负载管理器评估：验证任务随时间到达，需在共享 CPU 与内存限制下完成。测试显示孤立任务成功率不能预测流式表现：重负载下各系统均损失 13 至 23 分，中负载下两个孤立成功仅差 0.6 分的系统实际完成率分别为 54% 和 68%。串行调度会遗留 73% 的重负载任务未认领，而固定并发 2 的效果已接近最优 Agent 管理策略。
正文 · 原文
Abstract:Coding agents are moving from handling one request at a time to serving as persistent assistants to which users hand a stream of work: a build, a data-processing job and a bug fix may all be pending on the same machine at once. Such an agent must decide when to take on new tasks, how to delegate and overlap them, and how to stay within the machine's CPU and memory. Existing benchmarks cannot measure this, because they evaluate each task in isolation. We introduce \benchname{}, which evaluates an agent system as a workload manager: verified tasks arrive over time and must be delivered under shared CPU and memory limits, and the same tasks run one at a time serve as a per-system reference.
Across five model--harness pairs, isolated task success does not carry over evenly. On heavy workloads every system loses 13 to 23 points. On medium workloads, systems span 11 points in isolated success but 26 points in stream success, and two systems 0.6 points apart in isolation deliver 54\% and 68\% of the same tasks. Execution-policy interventions trace much of this gap to task intake: serial dispatch leaves 73\% of heavy-workload tasks unclaimed. Yet more elaborate management does not necessarily help. A fixed concurrency of two comes within four points of the best agent-managed policy on both medium and heavy workloads, and under every policy median CPU use stays below 30\% of the provisioned cores while tasks wait. These results identify workload management as a distinct capability for agents, and \benchname{} as a way to measure it.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.33178 [cs.AI] |
| (or arXiv:2610.33178v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.33178 arXiv-issued DOI via DataCite |
Submission history
From: Erle Zhu [view email]
[v1]
Fri, 23 Oct 2026 18:20:48 UTC (1,064 KB)
来源：arXiv cs.AI · arxiv.org