DeepResearch Bench: A Comprehensive Benchmark for Deep Research Agents

ICLR 2026 BENCHMARK

Mingxuan Du, Benfeng Xu, Chiwei Zhu, et al.

A comprehensive benchmark for evaluating deep research agents on complex scientific research tasks.

DeepResearch Bench II: Diagnosing Deep Research Agents via Rubrics from Expert Reports

PREPRINT 2026 BENCHMARK

Ruizhe Li, Mingxuan Du, Benfeng Xu, et al.

A diagnostic benchmark for deep research agents using expert-derived rubrics to evaluate report quality.

Wiki Live Challenge: Challenging Deep Research Agents with Expert-Level Wikipedia Articles

PREPRINT 2026 BENCHMARK

Shaohan Wang, Benfeng Xu, Licheng Zhang, et al.

A live benchmark leveraging Wikipedia Good Articles as expert-level references to evaluate deep research agents.

WildGraphBench: Benchmarking GraphRAG with Wild-Source Corpora

PREPRINT 2026 BENCHMARK GRAPHRAG

Pengyu Wang, Benfeng Xu, Licheng Zhang, et al.

A benchmark designed to assess GraphRAG performance in the wild using Wikipedia's external references as retrieval corpus, featuring 1,197 questions across 12 topics.