Paper 2602.04634
WideSeek-R1: Exploring Width Scaling for Broad Information Seeking via Multi-Agent Reinforcement Learning
- Published
- Feb 2026
- Research lab
- Independent
- Citations
- 6
- GitHub
- 4.5K stars
01 In brief
Summary
WIDESEEK-R1 explores width scaling for broad information seeking using a lead-agent–subagent framework trained via multi-agent reinforcement learning (MARL).
The system uses a shared LLM with isolated contexts and specialized tools, where the lead agent decomposes tasks and delegates subtasks to parallel subagents.
Training on a curated 20k dataset of broad information-seeking tasks, WIDESEEK-R1-4B achieves an item F1 score of 40.0% on the WideSearch benchmark, comparable to single-agent DeepSeek-R1-671B (41.3%) while using nearly 170x fewer parameters.
The method shows consistent performance gains as the number of parallel subagents increases, unlike depth scaling which plateaus.
Ablations confirm that joint optimization of both lead agent and subagents is necessary, and hybrid training data (wide and deep) yields the best results.
The model also surpasses larger multi-agent baselines on standard QA benchmarks, demonstrating that MARL enhances width scaling without compromising general reasoning.
02 From the paper
Abstract
Recent advancements in Large Language Models (LLMs) have largely focused on depth scaling, where a single agent solves long-horizon problems with multi-turn reasoning and tool use. However, as tasks grow broader, the key bottleneck shifts from individual competence to organizational capability. In this work, we explore a complementary dimension of width scaling with multi-agent systems to address broad information seeking. Existing multi-agent systems often rely on hand-crafted workflows and turn-taking interactions that fail to parallelize work effectively. To bridge this gap, we propose WideSeek-R1, a lead-agent-subagent framework trained via multi-agent reinforcement learning (MARL) to synergize scalable orchestration and parallel execution. By utilizing a shared LLM with isolated contexts and specialized tools, WideSeek-R1 jointly optimizes the lead agent and parallel subagents on a curated dataset of 20k broad information-seeking tasks. Extensive experiments show that WideSeek-R1-4B achieves an item F1 score of 40.0% on the WideSearch benchmark, which is comparable to the performance of single-agent DeepSeek-R1-671B. Furthermore, WideSeek-R1-4B exhibits consistent performance gains as the number of parallel subagents increases, highlighting the effectiveness of width scaling.