Long-Horizon-Terminal-Bench: Testing the Limits of Agents on Long-Horizon Terminal Tasks with Dense Reward-Based Grading

  • 类型:arxiv
  • 标识:2607.08964
  • 链接:https://arxiv.org/abs/2607.08964
  • 主分类:agent
  • 形态:benchmark
  • 被引:7
  • 被引来源:Semantic Scholar
  • S2被引:7
  • OpenAlex被引:0
  • 影响力被引:2
  • TLDR:This work introduces Long-Horizon-Terminal-Bench, a terminal benchmark of 46 long-horizon tasks spanning nine categories, including experiment reproduction, software engineering, multimodal analysis, interactive games, and scientific computing, and analyzes failure modes and error patterns to support future progress on long-horizon terminal agents.
  • OpenAlex ID:W7168172066
  • OpenAlex DOI:10.48550/arxiv.2607.08964
  • DOI:10.48550/arxiv.2607.08964
  • DOI来源:OpenAlex
  • 开放获取:green
  • 开放获取链接:https://doi.org/10.48550/arxiv.2607.08964
  • OpenAlex更新:2026-07-19
  • 副分类:evaluation
  • 待LLM分类:否
  • 标题中文:Long-Horizon-Terminal-Bench:基于密集奖励评分的 Agent 长程终端任务极限测试
  • TLDR中文:该工作提出 Long-Horizon-Terminal-Bench,一个涵盖九大类共 46 个长程任务的终端基准,包括实验复现、软件工程、多模态分析、交互游戏与科学计算,并分析失败模式与错误规律,以推动长程终端 Agent 的后续研究。
  • 来源文件
  • /inbox/tom/_candidates/2026-07-13-agent-rag-longcontext-candidates.json
  • [S2 enrich]
  • /inbox/tom/_candidates/2026-07-14-agent-rag-longcontext-candidates.json
  • /inbox/tom/_candidates/2026-07-14-rag-retrieval-reranking-candidates.json
  • [OpenAlex backfill]