Screen Tables and Substrate Scopes are information‑dense figures widely used in synthetic chemistry literature, yet automated extraction of structured reaction data from them remains challenging. We introduce RxnExBench, a multimodal benchmark for fine‑grained reaction‑level evaluation. Built from high‑impact chemistry journals, it contains 3,565 expert‑annotated reactions and 2,361 molecular structures. The benchmark adopts a unified schema with 15 reaction roles and 26 core fields, supporting downstream AI‑for‑Chemistry tasks such as retrosynthetic planning, yield prediction, and reaction condition recommendation. RxnExBench also provides a reasoning‑driven difficulty taxonomy, a modular annotation pipeline, and the Oracle‑MolRec protocol, which disentangles molecular recognition from higher‑level reasoning. Evaluation of nine state‑of‑the‑art vision‑language models shows that performance degrades substantially as schema complexity and reasoning demands increase. Under the Oracle‑MolRec setting, results reveal that accurate molecular recognition alone is insufficient; multimodal integration and chemistry‑aware reasoning remain the primary bottlenecks. All data and code are released to facilitate future research in chemistry‑aware multimodal understanding and AI‑for‑Chemistry dataset construction.
Review: RxnExBench offers a comprehensive benchmark for extracting structured chemical information, clearly exposing the current limitations of VLMs in chemical reasoning and guiding future model development.