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 |
|---|---|---|---|---|---|
| Mathematical Reasoning | OlympiadBench (test) | 40 | Jun 1, 2026 | ||
| Motion Sequence Interpolation | LaFAN1 | 40 | Feb 26, 2026 | ||
| Motion Sequence Interpolation | HumanML3D | 40 | Feb 26, 2026 | ||
| Deepfake Detection | FF++ video-level 8 (test) | 40 | Feb 26, 2026 | ||
| Reasoning Segmentation | ReasonSeg long (test) | 40 |
| Feb 26, 2026 |
| Reasoning Segmentation | ReasonSeg short (test) | 40 | Feb 26, 2026 |
|---|
| AI-generated image detection | CelebA-HQ v2 (test) | 40 | Feb 26, 2026 |
|---|
| Vehicle Re-Identification | RGBNT100 | 40 | May 12, 2026 |
|---|
| Land-cover classification | EuroSAT | 40 | Jun 23, 2026 |
|---|
| Audio-visual Navigation | Matterport3D Heard | 40 | Apr 8, 2026 |
|---|
| Audio-visual Navigation | Replica Heard | 40 | Apr 8, 2026 |
|---|
| Face Forgery Detection | CDF v1 | 40 | Mar 4, 2026 |
|---|
| Streaming Video Understanding | RVS-Movie | 40 | Jun 19, 2026 |
|---|
| Text Image Super-Resolution | Real-CE (val) | 40 | May 14, 2026 |
|---|
| Deepfake Detection | Celeb-DF (test) | 40 | Mar 17, 2026 |
|---|
| Text-to-Image Generation | DrawBench | 40 | Feb 26, 2026 |
|---|
| Compositional Hallucination Evaluation | OmniVCHall Average | 40 | Feb 26, 2026 |
|---|
| Compositional Hallucination Evaluation | OmniVCHall Generated | 40 | Feb 26, 2026 |
|---|
| Compositional Hallucination Evaluation | OmniVCHall Real | 40 | Feb 26, 2026 |
|---|
| Out-of-scope refusal | ScienceQA out-of-scope (test) | 40 | Feb 26, 2026 |
|---|
| Survival Analysis | TCGA-LUAD (test) | 40 | May 21, 2026 |
|---|
| 4D Generation | Consistent4D | 40 | Jun 2, 2026 |
|---|
| Human Preference Evaluation | ImageReward | 40 | Apr 21, 2026 |
|---|
| Counterfactual Reasoning | CVQA | 40 | Feb 27, 2026 |
|---|
| Art Style Erasure | Artist style and content prompts 5 groups SD v1.4 based (test) | 40 | Feb 26, 2026 |
|---|