AI Paper Atlas

Research map / Multi-agent systems and games / LLM agents

AgentFold: Long-Horizon Web Agents with Proactive Context Folding

Rui Ye, Zhongwang Zhang, Kuan Li, Huifeng Yin, Zhengwei Tao, Yida Zhao, Liangcai Su, Liwen Zhang, Zile Qiao, Xinyu Wang, Pengjun Xie, Fei Huang and 3 more

ICLR 2026 · Multi-agent systems and games › LLM agents · approach: Agentic benchmarks and orchestration

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes AgentFold, a web agent that proactively folds its interaction context at multiple scales to avoid context saturation on long-horizon information-seeking tasks.

Problem

LLM web agents on long-horizon information-seeking tasks suffer from context saturation with raw ReAct histories, while fixed per-step full-history summarization risks irreversible loss of critical details.

Approach

Instead of accumulating raw histories or summarizing everything at each step, the agent proactively folds its context (a learned context-management behavior) so that it keeps compact, reversible-in-spirit representations of past interactions while preserving critical details.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use proactive context folding (AgentFold) instead of append-only ReAct contextweb agent · context length at turn 100context tokens: over 84k tokens (92%) smallerImproved
scale the turn limit up to 256 with folded context (ablation)AgentFold-30B-A3B · BrowseCompaccuracyImproved
increase turn limit beyond 64 for append-only context (ablation)GLM-4.5-355B-A32B · BrowseCompaccuracyNo clear effect
train a 30B-A3B model with folding SFTQwen3-30B-A3B · BrowseCompaccuracy: 0.5 -> 36.2Improved
use AgentFold-30B-A3B in place of larger open modelsDeepSeek-V3.1-671B-A37B · BrowseComp-ZHaccuracy: 49.2 -> 47.3Mixed

Experimental setup

Baselines: ReAct-based agents fixed full-history summarization agents

Builds on: ReAct

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy