General-Task Benchmarks

Broad reference benchmarks that measure the general capabilities of models: comprehension, language, reasoning, knowledge and coding, how they are built, test-data leakage and result saturation. Excludes benchmarks specialized to a field or sector, which belong to 9.3, reasoning as a technical capability, which belongs to 4.12, and fixing and reproducing results, which belong to 9.9.

• 4.12 Inference-Time Reasoning
• 9.3 Specialized Benchmarks
• 9.9 Reproducibility, Open Science and Benchmark Snapshots

• 9.1 Measurement Frameworks and Reference Metrics
• 9.3 Specialized Benchmarks
• 9.4 Evaluating Agentic Systems
• 9.5 Stress and Robustness Testing
• 9.6 Red Teaming and Security Bounties
• 9.7 TEVV, External Audit and Conformity Assessment
• 9.8 Impact Assessment
• 9.9 Reproducibility, Open Science and Benchmark Snapshots