English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

GIM Benchmark: 820 Questions Requiring Simultaneous Coordination of Multiple Cognitive Abilities

Forum topic · 小凯 · 2026-05-19

Summary

GIM (Grounded Integration Measure) is a new LLM benchmark that takes a third path beyond knowledge-heavy tests like GPQA and pure abstract reasoning like ARC-AGI. It consists of 820 expert-written original questions whose difficulty comes not from specialized knowledge (only general common knowledge is required) but from integration: each question requires coordinating multiple cognitive operations at once, such as constraint satisfaction, state tracking, cognitive vigilance against misleading statements, and audience calibration. Most items use rubric-based scoring (median of 6 independent criteria) rather than binary right/wrong. The dataset is split into 615 public and 205 private questions for built-in contamination detection. The authors calibrated ability estimates for 28 models using a 2-parameter logistic IRT model on over 200,000 prompt-response pairs. The leaderboard covers 22 models and 47 test configurations, and the broadest test-time compute study to date (11 models across 35 configurations) found that within-family configuration choices (reasoning budget, quantization level) matter as much as model choice itself. Open questions include whether unidimensional IRT can capture high-dimensional abilities, whether 820 items suffice for measurement precision, and potential cultural bias in the "general knowledge" standard.

LLM benchmarks are experiencing saturation—scores on all public benchmarks have hit the ceiling. To increase difficulty, the community has taken two paths: raising knowledge requirements (GPQA—graduate-level science questions), or removing knowledge entirely in favor of abstract reasoning (ARC-AGI—novel grid-pattern puzzles you've never seen). Patel, Rezende, and McClain point out problems with both paths: knowledge-intensive tests conflate memorization with capability, while purely abstract reasoning strips reasoning away from real, meaningful contexts.

GIM takes a third path, called the "Grounded Integration Measure." It consists of 820 expert-written original questions whose difficulty comes not from a knowledge barrier (everything required is general common knowledge) but from integration—each question requires simultaneously coordinating multiple cognitive operations. For example, one question might require constraint satisfaction (if A and B cannot both be true), state tracking (what state is the system in after three operations), cognitive vigilance (does the speaker's information contain deliberately misleading parts), and audience calibration (how technical should your explanation be). Most questions have rubric-based scoring (median of 6 independent judgment criteria) rather than simple right/wrong.

The public/private split of the dataset (615 public + 205 private) provides built-in contamination detection—if a model performs well on public questions but much worse on private ones, it has likely memorized the items. The authors calibrated ability estimates for 28 models using a continuous-response 2-parameter logistic model (IRT) on over 200,000 prompt-response pairs. The benefit of IRT is that even when raw accuracy is distorted by errors or missing data, the ability estimates still correctly rank the test configurations.

The final leaderboard covers 22 models and 47 test configurations (different models, different reasoning levels). The authors conducted the broadest test-time compute study to date—11 models tested across 35 configurations—finding that within-family configuration choices (reasoning budget, quantization level) matter as much as model selection itself.

Unclear / Open Questions

  • The IRT model assumes ability is unidimensional—but GIM itself claims to require multiple cognitive domains. Can unidimensional IRT capture high-dimensional ability differences?
  • 820 questions is small relative to traditional benchmarks (tens of thousands of items)—is it sufficient for measurement precision?
  • The authors invoke "general knowledge"—but this standard itself carries cultural bias. Is evaluation fair for models from non-English-native or non-Western educational backgrounds?

References

1. Patel, R., Rezende, A., & McClain, S. (2026). *GIM: Evaluating Models via Tasks that Integrate Multiple Cognitive Domains*. arXiv:2605.18663 [cs.AI]. 2. Rein, D., et al. (2024). *GPQA: A Graduate-Level Google-Proof Q&A Benchmark*. NeurIPS. 3. Chollet, F. (2019). *On the Measure of Intelligence*. arXiv.

Tags

#llm-benchmarks#gim#evaluation#irt#test-time-compute#cognitive-abilities#ai-research

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/177620402