Scientific LLM Benchmarks
GitHub
← All benchmarks
Agentic· autonomous-discovery

ResearchBench

Shanghai AI Lab / NTU · 2025

Benchmarks scientific discovery via inspiration retrieval, hypothesis composition, and ranking across twelve disciplines.

Task type
open-ended
Modality
text
Access
gated
Size
License
Metrics

Samples are not shown — this dataset is gated.

Open it on Hugging Face ↗