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 |
|---|---|---|---|---|---|
| Gallbladder Cancer Detection | Gallbladder Cancer Ultrasound (GBC-US) (5-fold cross-val) | 32 | Feb 26, 2026 | ||
| Online Mapping | nuScenes (val) | 32 | Apr 8, 2026 | ||
| Vision-Language Navigation | HA-VLN Unseen (val) | 32 | May 14, 2026 | ||
| 3D Human Interaction Reconstruction | Hi4D | 32 | Jun 16, 2026 | ||
| Video Repetition Counting | UCFRep (test) |
| 32 |
| Feb 26, 2026 |
| Causal Question Answering | CausalProbe-H | 32 | Feb 26, 2026 |
|---|
| Causal Question Answering | CausalProbe-E | 32 | Feb 26, 2026 |
|---|
| Causal Question Answering | CausalNet | 32 | Feb 26, 2026 |
|---|
| Causal Question Answering | e-CARE | 32 | Feb 26, 2026 |
|---|
| Causal Question Answering | COPA | 32 | Feb 26, 2026 |
|---|
| Optical Flow | Spring (test) | 32 | Jun 30, 2026 |
|---|
| Safety Alignment | AdvBench | 32 | Mar 17, 2026 |
|---|
| Pathology report generation | PathText BRCA (test) | 32 | Mar 5, 2026 |
|---|
| Semantic Segmentation | Pascal Blended 1/4 augmented (train) | 32 | Feb 26, 2026 |
|---|
| Semantic Segmentation | Pascal Blended augmented (1/8 train) | 32 | Feb 26, 2026 |
|---|
| RGB-T Tracking | RGBT210 (test) | 32 | Feb 26, 2026 |
|---|
| Trivia QA | Trivia QA | 32 | Feb 26, 2026 |
|---|
| Downstream Task | 11 Downstream Tasks Aggregate | 32 | Feb 26, 2026 |
|---|
| Hepatocellular carcinoma (HCC) detection | Hepatocellular carcinoma (HCC) | 31 | Jul 7, 2026 |
|---|
| Image Classification | CIFAR-10 AutoAttack v1 (AA) | 31 | Jul 7, 2026 |
|---|
| Multimodal Perception and Reasoning | MME | 31 | Jul 7, 2026 |
|---|
| Few-shot classification | FSC-89 → NSynth-100 | 31 | Jul 8, 2026 |
|---|
| Visual Odometry | TUM-VI (room sequences) | 31 | Jul 2, 2026 |
|---|
| Image-text retrieval | Flickr8k (test) | 31 | Jun 30, 2026 |
|---|
| Multi-turn Agent Task | Search-QA | 31 | Jul 1, 2026 |
|---|