Latest update — July 2026. SciConBench is a live benchmark updated monthly with newly released CDSR reviews. This release expands the benchmark from 9.11K to 9.17K samples and replaces superseded reviews with their latest editions through July 2026.
SciConBench is a large-scale, live benchmark for evaluating AI agents on open-domain scientific conclusion synthesis: the long-horizon task of retrieving relevant evidence from the open web, assessing evidence… See the full description on the dataset page:
https://huggingface.co/datasets/hayoungjung/SciConBench.
