Research
Benchmarks
Compare state-of-the-art methods across the tasks and datasets used to measure AI progress.
309,131 benchmarksPage 595 of 12,366
Benchmarks
| Task | Dataset | Domain | Trend | Results | Last update |
|---|---|---|---|---|---|
| Multi-modal Question Answering | MedXpertQA-MM | 38 | Apr 13, 2026 | ||
| Multimodal Understanding | MMBench en (dev) | 38 | Feb 27, 2026 | ||
| Speech-Preserving Facial Expression Manipulation | MEAD (cross-identity) | 38 | Feb 26, 2026 | ||
| Object Detection | VEDAI | 38 | Mar 10, 2026 | ||
| Streaming Video Understanding | OVOBench Realtime | 38 | Jun 25, 2026 | ||
| Class-to-image generation | ImageNet 256x256 | 38 | May 8, 2026 | ||
| Visual Question Answering | MME | 38 | Jun 11, 2026 | ||
| Super-Resolution | RealSR x3 (test) | 38 | Feb 26, 2026 | ||
| Point Tracking | DAVIS | 38 | Feb 26, 2026 | ||
| Brain disorder classification | ADNI (Five-fold cross-validation) | 38 | Jun 4, 2026 | ||
| Context-Aware Image Retrieval | DISBench Inter-Event | 38 | Jun 4, 2026 | ||
| Jailbreak Detection | Average of six attacks | 38 | Apr 3, 2026 | ||
| Offline Reinforcement Learning | Maze2D medium | 38 | Feb 26, 2026 | ||
| Offline Reinforcement Learning | Maze2D umaze | 38 | Feb 26, 2026 | ||
| Malicious Prompt Refusal | HarmBench | 38 | Jun 23, 2026 | ||
| Jailbreak Attack | AdvBench GPT-3.5-turbo 1.0 (test) | 38 | Mar 25, 2026 | ||
| Malware Classification | Mal-Net | 38 | Feb 26, 2026 | ||
| Long-context understanding | RULER 32K | 38 | May 19, 2026 | ||
| Long-context Question Answering | DetectiveQA-En | 38 | May 8, 2026 | ||
| Long-context Question Answering | DetectiveQA-Zh | 38 | May 8, 2026 | ||
| Long narrative understanding QA | NoCha | 38 | May 8, 2026 | ||
| Code Generation | BigCodeBench-Completion Hard | 38 | Feb 26, 2026 | ||
| Information Retrieval | Scientific QA Base setting | 38 | Feb 26, 2026 | ||
| Question Answering | Scientific QA Base setting | 38 | Feb 26, 2026 | ||
| Binary Classification | Diabetes | 38 | May 13, 2026 |