Research
Benchmarks
Compare state-of-the-art methods across the tasks and datasets used to measure AI progress.
309,131 benchmarksPage 153 of 12,366
Benchmarks
| Task | Dataset | Domain | Trend | Results | Last update |
|---|---|---|---|---|---|
| Long-video understanding | Video-MME | 90 | Jun 11, 2026 | ||
| Image Classification | DTD (Describable Textures Dataset) | 90 | Jun 12, 2026 | ||
| Robustness Verification | Iris dataset (test) | 90 | Mar 4, 2026 | ||
| Multimodal Abstractive Summarization | MM-Sum mid-high-resource | 90 | Feb 27, 2026 | ||
| Longest Functional Common Subsequence (LFCS) | SMALL literature instances | 90 | Feb 27, 2026 | ||
| Indirect Prompt Injection Attack | IPI-3k | 90 | Feb 26, 2026 | ||
| OOD detection | MVTec Metal Nut Dissimilar | 90 | Feb 27, 2026 | ||
| OOD Detection | MVTec Pill (Dissimilar) | 90 | Feb 27, 2026 | ||
| OOD Detection | MVTec Pill (Similar) | 90 | Feb 27, 2026 | ||
| Multi-turn Dialogue Evaluation | MT-Bench-zh | 90 | Feb 26, 2026 | ||
| Dialogue Alignment Evaluation | AlignBench | 90 | Feb 26, 2026 | ||
| Inference Efficiency | Natural Questions (NQ) | 90 | Mar 23, 2026 | ||
| Formal Conformity Evaluation | Songci 20 Cipai and 6 themes | 90 | Feb 26, 2026 | ||
| Mathematical Reasoning | Olympiad | 90 | Mar 4, 2026 | ||
| Black-box Optimization | CEC 30D 2017 | 90 | Feb 27, 2026 | ||
| Visual Classification | 8 Vision Tasks (SUN397, Stanford Cars, RESISC45, EuroSAT, SVHN, GTSRB, MNIST, DTD) | 90 | Jun 30, 2026 | ||
| Physical Reasoning | PIQA | 90 | May 20, 2026 | ||
| Readmission Prediction | MIMIC-IV | 90 | Jun 24, 2026 | ||
| Instruction Following | Natural Instructions (test) | 90 | Feb 26, 2026 | ||
| Click-Through Rate Prediction | AutoML | 90 | Feb 26, 2026 | ||
| Backdoor Attack Robustness | OpenSARShip (test) | 90 | Feb 26, 2026 | ||
| Backdoor Attack Robustness | MSTAR (test) | 90 | Feb 26, 2026 | ||
| Safety evaluation | Sorry-Bench | 90 | Feb 26, 2026 | ||
| Machine Unlearning | MUSE Books | 90 | Jun 11, 2026 | ||
| Safety Evaluation | WildJailbreak | 90 | Jun 16, 2026 |