SGI-Bench: Scientist-Aligned LLM Benchmark Across 10 Disciplines

SGI-Bench is a benchmark for evaluating Scientific General Intelligence (SGI) of LLMs across the full inquiry cycle. It spans 10 scientific disciplines and contains more than 1,000 expert-curated samples inspired by Science's 125 Big Questions. The dataset was created by InternScience and was last updated on June 2, 2026.

Use Cases

• Benchmarking LLM performance on scientific deliberation tasks based on the described inquiry cycle
• Evaluating model conception of scientific hypotheses across 10 disciplines
• Testing agentic action planning for scientific workflows mentioned in the description
• Assessing perception and interpretation of scientific problems inspired by major questions

Strengths

• More than 1,000 expert-curated samples provide a substantial test set
• Benchmark spans 10 distinct scientific disciplines for broad coverage
• Evaluation framework is aligned with scientist workflows across the full inquiry cycle
• Samples are inspired by Science's 125 Big Questions, grounding them in recognized scientific challenges

Limitations

• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment
• Description metadata is limited; actual data quality requires manual inspection after download

Provenance

Source

InternScience

Collection Method

Expert-curated samples inspired by Science's 125 Big Questions

Freshness

Last updated 2026-06-02 12:03:26; freshness should be verified