Research
Benchmarks
Compare state-of-the-art methods across the tasks and datasets used to measure AI progress.
309,131 benchmarksPage 1585 of 12,366
Benchmarks
| Task | Dataset | Domain | Trend | Results | Last update |
|---|---|---|---|---|---|
| Selective Prediction | WMT de 19 | 20 | Apr 7, 2026 | ||
| Selective Prediction | WMT fr 14 | 20 | Apr 7, 2026 | ||
| Selective Prediction | WMT de 14 | 20 | Apr 7, 2026 | ||
| Selective Prediction | GSM8K | 20 | Apr 7, 2026 | ||
| Selective Prediction | TruthfulQA | 20 | Apr 7, 2026 | ||
| Selective Prediction | MMLU | 20 | Apr 7, 2026 | ||
| Selective Prediction | CoQA | 20 | Apr 7, 2026 | ||
| Selective Prediction | BaBi | 20 | Apr 7, 2026 | ||
| Classification | ACS dataset I=[20%, 40%] (test) | 20 | Apr 7, 2026 | ||
| Encrypted Sparse Matrix-Vector Multiplication | SuiteSparse Matrix Collection 20 matrices with top gains | 20 | Apr 7, 2026 | ||
| Question Answering | SimpleQA | 20 | May 21, 2026 | ||
| Membership Inference Attack | 4 natural language datasets (aggregated) | 20 | Apr 6, 2026 | ||
| Membership Inference Attack | 4 natural language datasets (aggregated) (test) | 20 | Apr 6, 2026 | ||
| Reward Modeling Evaluation | RewardBench2 (test) | 20 | Apr 6, 2026 | ||
| Web Browsing | BrowseComp+ (test) | 20 | Apr 6, 2026 | ||
| Agentic Tool-use | AppWorld (Challenge) | 20 | Apr 6, 2026 | ||
| Agentic Tool-use | AppWorld Normal | 20 | Apr 6, 2026 | ||
| Image Classification | Infrared-COCO | 20 | Apr 6, 2026 | ||
| Mapping Quality | BundleFusion | 20 | Apr 6, 2026 | ||
| Table Structure Recognition | MUSTARD | 20 | Apr 6, 2026 | ||
| Mathematical Reasoning | OlympiadBench | 20 | Apr 7, 2026 | ||
| Declouding | RICE2 (test) | 20 | Apr 6, 2026 | ||
| Dynamic 3D Reconstruction | coffee martini | 20 | Apr 6, 2026 | ||
| Dynamic 3D Reconstruction | taekwondo | 20 | Apr 6, 2026 | ||
| Dynamic 3D Reconstruction | vrheadset | 20 | Apr 6, 2026 |