---
id: 20261022-T0-13
title: "网页Agent总因预算和封禁失败？新决策模型把成本与风险纳入强化学习"
title_en: "New Web Agent RL Model Factors Request Budgets and Failure Risk Into Decisions"
url: https://ai.daily.yangsir.net/daily/20261022-T0-13
issue_date: 2026-10-22
publish_date: 2026-10-21T04:00:00.000Z
category: research
source_name: "arXiv cs.LG (ML)"
source_url: https://arxiv.org/abs/2610.23857
---

# 网页Agent总因预算和封禁失败？新决策模型把成本与风险纳入强化学习

针对网页环境中的智能体交互常受请求预算、延迟、反爬限制和操作失败风险约束的问题，一项新研究提出在强化学习决策中同时建模多类成本与失败风险，而不是只优化任务成功率。原摘要指出，仅追求成功率的策略在实际部署中往往不可用。该工作为网页Agent在真实预算和风控条件下如何权衡成功率与成本提供了决策框架，可用于需要控制请求量和失败代价的自动化网页任务。

## English Version

**New Web Agent RL Model Factors Request Budgets and Failure Risk Into Decisions**

A new arXiv paper addresses web agents that operate under request budgets, latency, anti-crawling restrictions and operational failure risk. The authors argue that strategies optimizing only task success rates often become unusable in practice, and propose an agentic reinforcement learning decision model that accounts for multiple costs and failure risk. The work offers a framework for trading off success rate against request volume and failure penalties in real-world automated web tasks.

---

**来源**：[arXiv cs.LG (ML)](https://arxiv.org/abs/2610.23857)

**详情页**：https://ai.daily.yangsir.net/daily/20261022-T0-13

---

*智语观潮 · Daily — https://ai.daily.yangsir.net/llms.txt*