Research
Benchmarks
Compare state-of-the-art methods across the tasks and datasets used to measure AI progress.
309,131 benchmarksPage 2349 of 12,366
Benchmarks
| Task | Dataset | Domain | Trend | Results | Last update |
|---|---|---|---|---|---|
| Image Editing | GRADE (test) | 15 | Jul 3, 2026 | ||
| Linear Regression In-context Learning | Garg Linear Regression d=10 | 15 | Jul 3, 2026 | ||
| Semantic Segmentation | Cityscapes (val) | 15 | Jul 3, 2026 | ||
| Underwater Salient Object Detection | RMAS (test) | 15 | Jul 3, 2026 | ||
| Underwater Salient Object Detection | MAS3K (test) | 15 | Jul 3, 2026 | ||
| LLM root-cause identification | LLM root-cause identification dataset n=30 per model | 15 | Jul 3, 2026 | ||
| Vulnerability Reproduction | CyberGym (evaluation) | 15 | Jul 3, 2026 | ||
| Preference Alignment | U10 (held-out evaluation set) | 15 | Jul 3, 2026 | ||
| Text-to-SQL | Spider Out-of-Domain (dev) | 15 | Jul 3, 2026 | ||
| General Utility Evaluation | MT-Bench | 15 | Jul 3, 2026 | ||
| Fatal Error | OpenAgent Tier-4 | 15 | Jul 2, 2026 | ||
| Logic Inversion | OpenAgent Tier-3 | 15 | Jul 2, 2026 | ||
| Tool Redirection | OpenAgent Tier-2 | 15 | Jul 2, 2026 | ||
| Semantic Trap | OpenAgent Tier-1 | 15 | Jul 2, 2026 | ||
| Reward Modeling | OPENRM-LIB v1 (test) | 15 | Jul 2, 2026 | ||
| Speaker Verification | VoxCeleb1 Black-box FAKEBOB | 15 | Jul 2, 2026 | ||
| Speaker Verification | VoxCeleb1 Targeted PGD-ℓ∞ | 15 | Jul 2, 2026 | ||
| Speaker Verification | VoxCeleb1 Targeted PGD-ℓ2 | 15 | Jul 2, 2026 | ||
| Speaker Verification | VoxCeleb1 Targeted MI-FGSM | 15 | Jul 2, 2026 | ||
| Speaker Verification | VoxCeleb1 Genuine | 15 | Jul 2, 2026 | ||
| Generalized Assignment Problem | ORLIB | 15 | Jul 2, 2026 | ||
| Scientific Reasoning | GPQA (test) | 15 | Jul 2, 2026 | ||
| Video-to-Text Retrieval | CapERA | 15 | Jul 2, 2026 | ||
| Video-to-Text Retrieval | DVTMD | 15 | Jul 2, 2026 | ||
| Defense against adversarial prompts | P4D (test) | 15 | Jul 2, 2026 |