Reasoning Benchmarks
MMLU, ARC, HellaSwag, and other reasoning evaluations tracking logical inference and knowledge retrieval across model families.
AI Benchmark Index
Performance benchmarks across reasoning, coding, math, and multimodal tasks for frontier and open-source models. Automated weekly updates from public leaderboards.
View Dataset on GitHub →What This Index Covers
Coding Benchmarks
HumanEval, MBPP, SWE-bench, and code generation evaluations measuring programming capability across languages and complexity levels.
Math Benchmarks
GSM8K, MATH, and competition-level mathematics evaluations tracking numerical reasoning and problem-solving performance.
Multimodal Benchmarks
Vision-language evaluations, image understanding, and cross-modal reasoning benchmarks for multimodal AI systems.
Methodology
Data is collected weekly via automated pipelines from public leaderboards, model cards, academic papers, and reproducible evaluation frameworks. All collection scripts are transparent and auditable.
40+
Models Tracked
Weekly
Update Frequency
100%
Open Source