Research
Benchmarks
Compare state-of-the-art methods across the tasks and datasets used to measure AI progress.
309,131 benchmarksPage 316 of 12,366
Benchmarks
| Task | Dataset | Domain | Trend | Results | Last update |
|---|---|---|---|---|---|
| Density Estimation | Banana-shaped distributions scenario Banana | 56 | May 14, 2026 | ||
| Density Estimation | Gaussian mixture distributions pre-training family F GMDF+ | 56 | May 14, 2026 | ||
| Density Estimation | Gaussian mixture distributions pre-training family F GMDF | 56 | May 14, 2026 | ||
| Data Contamination Detection | Seen vs. Unseen Datasets (dataset-level) | 56 | May 13, 2026 | ||
| Language Modeling | WikiText-103 v1 (test) | 56 | May 13, 2026 | ||
| Multilingual Evaluation | TAXI1500 | 56 | May 13, 2026 | ||
| Multilingual Evaluation | SIB200 | 56 | May 13, 2026 | ||
| Membership Auditing | Math Reasoning Tasks | 56 | May 12, 2026 | ||
| Reasoning | ARC Challenge | 56 | Jun 23, 2026 | ||
| Scientific Reasoning | SciKnowEval | 56 | Jul 3, 2026 | ||
| Claim Verification | Med Claim | 56 | May 8, 2026 | ||
| Action Category Classification | CLIP | 56 | May 8, 2026 | ||
| Dynamic Link Prediction | Enron (transductive) | 56 | Jun 4, 2026 | ||
| Safety Evaluation | JailbreakBench (JBB) (test) | 56 | May 7, 2026 | ||
| Safety Evaluation | Refusal Signal Score | 56 | May 7, 2026 | ||
| Safety Evaluation | SecureBreak | 56 | May 7, 2026 | ||
| Safety Evaluation | Llama-Guard 3-8B | 56 | May 7, 2026 | ||
| User response prediction | WorldValuesBench (held-out users) | 56 | May 4, 2026 | ||
| Response Prediction | WorldValuesBench (held-out) | 56 | May 4, 2026 | ||
| Ordinal response prediction | Synthetic users N=20,000 (test) | 56 | May 4, 2026 | ||
| Ordinal Prediction | WorldValuesBench (held-out users) | 56 | May 4, 2026 | ||
| Adaptive Querying | Synthetic users N = 20,000 (test) | 56 | May 4, 2026 | ||
| Question Answering | ARC Challenge | 56 | Jun 18, 2026 | ||
| Sentence Similarity | SICK | 56 | Apr 28, 2026 | ||
| Harmfulness Evaluation | DirectHarm | 56 | Apr 24, 2026 |