Loading the SOTA2 catalog…
SOTA2 Research · benchmarks
Compare state-of-the-art methods across the tasks and datasets used to measure AI progress.
| Task | Dataset | Domain | Trend | Results | Last Update |
|---|---|---|---|---|---|
| Combinatorial Reasoning | Graph Coloring | 30 | May 26, 2026 | ||
| Image Classification | CIFAR-100 | 30 | Mar 4, 2026 | ||
| Retrosynthesis | USPTO-50k Without Reaction Type | 30 | Mar 4, 2026 | ||
| Stability Prediction | Matbench-Discovery unique structure prototypes | 30 | May 21, 2026 | ||
| Point Cloud Completion | Rabbit point cloud | 30 | Mar 17, 2026 |
| Point Cloud Completion | Mario point cloud | 30 | Mar 17, 2026 |
|---|
| Point Cloud Completion | Frog point cloud | 30 | Mar 17, 2026 |
|---|
| Remote Sensing Image Dehazing (RSI Dehazing) | RICE1 | 30 | Apr 6, 2026 |
|---|
| Jailbreak Robustness | StrongReject | 30 | May 20, 2026 |
|---|
| Copy | Copy OOD lengths: 2x, 4x, 8x, 16x, 32x, 64x | 30 | Mar 4, 2026 |
|---|
| MQMTAR | MQMTAR OOD lengths 2x 4x 16x 64x 256x 1024x | 30 | Mar 4, 2026 |
|---|
| Multiple Instance Learning Classification | TIGER | 30 | Jun 5, 2026 |
|---|
| Multiple Instance Learning Classification | FOX | 30 | Jun 5, 2026 |
|---|
| Mobility Prediction | SPB | 30 | Mar 4, 2026 |
|---|
| Mobility Prediction | MOW | 30 | Mar 4, 2026 |
|---|
| Mobility Prediction | MAO | 30 | Mar 4, 2026 |
|---|
| Mobility Prediction | KLP Foursquare Gowalla (test) | 30 | Mar 4, 2026 |
|---|
| Mobility Prediction | JKT Foursquare Gowalla (test) | 30 | Mar 4, 2026 |
|---|
| Identification of inactive attention heads | WinoGrande | 30 | Mar 4, 2026 |
|---|
| Document Question Answering | LongDocURL | 30 | Apr 13, 2026 |
|---|
| OOD detection | MVTec Metal Nut - Similar artifacts | 30 | Feb 27, 2026 |
|---|
| OOD Detection | MVTec Metal Nut (Similar) | 30 | Feb 27, 2026 |
|---|
| Video Super-Resolution | Vid4 Y (test) | 30 | Feb 27, 2026 |
|---|
| Multimodal Classification | Kinetics-Sounds (test) | 30 | Apr 8, 2026 |
|---|
| Handwriting Retrieval | Handwriting Spanish synthetic disjoint fonts (Out-of-Domain (OOD)) | 30 | Feb 27, 2026 |
|---|