Research
Benchmarks
Compare state-of-the-art methods across the tasks and datasets used to measure AI progress.
309,131 benchmarksPage 1230 of 12,366
Benchmarks
| Task | Dataset | Domain | Trend | Results | Last update |
|---|---|---|---|---|---|
| Multi-path Speculative Decoding | Held-out (test) | 24 | Feb 26, 2026 | ||
| General Utility | Utility Benchmark (test) | 24 | Feb 26, 2026 | ||
| Safety Overrefusal Evaluation | Overrefusal Evaluation Suite (test) | 24 | Feb 26, 2026 | ||
| Jailbreaking Defense | Jailbreaking Evaluation Suite (test) | 24 | Feb 26, 2026 | ||
| Goal prediction | Pedestrian Dataset campus navigation (val) | 24 | Feb 26, 2026 | ||
| Preference Learning | Toy dataset 50% label noise (test) | 24 | Feb 26, 2026 | ||
| Mathematical Reasoning | OlympiadBench | 24 | Feb 26, 2026 | ||
| Mathematical Reasoning | MATH500 | 24 | Feb 26, 2026 | ||
| Fingerprint Robustness Evaluation | Prominent Deployment Scenarios Robustness Evaluation 1.0 | 24 | Feb 26, 2026 | ||
| Causal Discovery | Synthetic Data |V|=5 | 24 | Mar 6, 2026 | ||
| Regression | Openml_586 | 24 | Mar 12, 2026 | ||
| Regression | Airfoil UCIrvine | 24 | Mar 12, 2026 | ||
| Question Answering | 2WikiQA (test) | 24 | Jun 9, 2026 | ||
| Regression | Scenario IS2 | 24 | Feb 26, 2026 | ||
| Classification | Scenario IS1 | 24 | Feb 26, 2026 | ||
| Check-in Prediction | NYC Target from CHI & SF | 24 | Feb 26, 2026 | ||
| Sequential Recommendation | Amazon Sport | 24 | Feb 26, 2026 | ||
| Mathematical Reasoning | Average Six Benchmarks | 24 | Feb 26, 2026 | ||
| Super-Resolution | TNO | 24 | Feb 26, 2026 | ||
| Open-Domain Question Answering | LoCoMo Open-Domain (test) | 24 | Feb 26, 2026 | ||
| Temporal Question Answering | LoCoMo Temporal (test) | 24 | Feb 26, 2026 | ||
| Single-Hop Question Answering | LoCoMo Single-Hop (test) | 24 | Feb 26, 2026 | ||
| Multi-Hop Question Answering | LoCoMo Multi-Hop (test) | 24 | Feb 26, 2026 | ||
| Jailbreak Robustness | JBB-Behaviors (test) | 24 | Feb 26, 2026 | ||
| Multi-hop Question Answering | InfoQA Synthetic Multi-hop QA 1-hop | 24 | Feb 26, 2026 |