arXiv cs.AI· Michael Ofengenden, Maksym Andriushchenko·· 9 天前AI 评分41
AgentTime：AI 智能体能否预估并控制自身运行时长？
AgentTime: Can Agents Estimate and Control Their Own Runtime?
AI 导读
研究者提出 AgentTime 基准，用 222 个来自 18 个来源的任务测试智能体能否按指定时长工作、预测运行时间并事后估算耗时。在时长遵循实验中，Claude Code 中的 Fable 5.1 与要求时长的典型偏差达 2.9 倍，而 Codex 中的 GPT-6 Astra 仅为 1.2 倍；158 次可分类的 Astra 运行中有 14 次在看似完成后显式休眠。
正文 · 原文
Abstract:An essential control of AI agents is their ability to manage runtime. This ability requires a sense of time-awareness, to predict and estimate wall-clock time and to control their own actions. Prior work has focused on time-awareness, but duration-following and control in native agent harnesses remain unexplored. We present AgentTime, a benchmark for testing whether agents can work for a requested duration, predict their runtime, and estimate elapsed time afterward. It comprises 222 tasks from 18 sources spanning coding, computer use, agentic work, and automated research. Duration-following experiments append a single instruction specifying how long to work, with requests ranging from about a minute to multiple days. Accuracy on these instructions varies substantially: Fable 5.1 in Claude Code deviates from requested runtimes by a typical factor of 2.9$\times$, compared with only 1.2$\times$ for GPT-6 Astra in Codex. However, matching the requested runtime does not, by itself, establish continued work on the task. Among 158 reviewed Astra runs with classifiable transcripts, 14 explicitly slept after appearing to finish. In forecasting experiments, predictions tend to overestimate natural runtimes. In retrospective experiments, removing temporal information more than doubles deviation for Sol and Astra and nearly doubles it for Fable. An agent's ability to complete a task does not guarantee that it can control its own time or work for the whole requested duration. For agents to run reliably, safely, and autonomously over long horizons, we require the evaluation of both.
| Comments: | Website: this https URL Code: this https URL |
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.09944 [cs.AI] |
| (or arXiv:2610.09944v2 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.09944 arXiv-issued DOI via DataCite |
Submission history
From: Michael Ofengenden [view email]
[v1]
Wed, 7 Oct 2026 12:26:24 UTC (1,219 KB)
[v2]
Thu, 8 Oct 2026 09:06:36 UTC (1,219 KB)
来源：arXiv cs.AI · arxiv.org