AgentLongBench: A Controllable Long Benchmark For Long-Contexts Agents via Environment Rollouts

Shicheng Fang, Yuxin Wang, Xiaoran Liu (SII), Jiahao Lu, Chuanyuan Tan, Xinchi Chen, Yining Zheng. Xuanjing Huang, Xipeng Qiu

AgentLongBench: A Controllable Long Benchmark For Long-Contexts Agents via Environment Rollouts: 6 upvotes on Hugging Face Daily Papers, #14 of 45 papers on 2026-01-30. Day-by-day upvote history. It lost 16 votes when the Hub removed votes in bulk.

The evolution of Large Language Models (LLMs) into autonomous agents necessitates the management of extensive, dynamic contexts. Current benchmarks, however, remain largely static, relying on passive retrieval tasks that fail to simulate the complexities of agent-environment interaction, such as non-linear reasoning and iterative feedback. To address this, we introduce AgentLongBench, which evaluates agents through simulated environment rollouts based on Lateral Thinking Puzzles. This framework generates rigorous interaction trajectories across knowledge-intensive and knowledge-free scenarios. Experiments with state-of-the-art models and memory systems (32K to 4M tokens) expose a critical weakness: while adept at static retrieval, agents struggle with the dynamic information synthesis essential for workflows. Our analysis indicates that this degradation is driven by the minimum number of tokens required to resolve a query. This factor explains why the high information density inherent in massive tool responses poses a significantly greater challenge than the memory fragmentation typical of long-turn dialogues.

Paper page on Hugging Face · arXiv

Data: hysts-bot-data/daily-papers-stats and the Daily Papers API. Open data: tardellirs/paper-pulse-data. Sister project: Model Pulse, the download history of every model on the Hub.