| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| SugarCrepe | Overall Accuracy95.2 | 95 | 18d ago | ||
| VALSE | BEiTScore | Average Score89.2 | 65 | 29d ago | |
| VL-Checklist | SGVL | Attribute Score81.8 | 47 | 1mo ago | |
| Winoground | Qwen2.5-VL-7B + SSL-R1 | Group Score41.25 | 33 | 18d ago | |
| SugarCrepe++ | CLIPS (ViT-L/14) | Average Performance75.2 | 25 | 2mo ago | |
| Compositional Reasoning Suite Aggregated | LLaVA-7B† | Sugarcrepe Score93.1 | 23 | 3mo ago | |
| CREPE | ReasonCLIP + READ | Accuracy44.9 | 21 | 29d ago | |
| WhatsUp | ReasonCLIP | Accuracy51.3 | 21 | 29d ago | |
| WebAggregatorQA | WebAggregator | Level-1 Score70.8 | 21 | 2mo ago | |
| GQA | Accuracy (GQA Compositional Reasoning)62 | 18 | 1mo ago | ||
| GenAI-Bench (test) | CDG | Spatial Score83.79 | 18 | 4mo ago | |
| ARO | CE-CLIP+ | Relation Score83.6 | 17 | 1mo ago | |
| BiVLC | TinyLLaVA | Accuracy94.6 | 16 | 18d ago | |
| VisMin Easy (test) | Qwen3-VL | Text Score94.88 | 16 | 1mo ago | |
| MMStar | GPT-4o | Accuracy64.7 | 16 | 3mo ago | |
| Lego Puzzles | GPT-4o-0513 | Accuracy57.7 | 13 | 18d ago | |
| BLINK | GPT-4o | Accuracy68 | 12 | 3mo ago | |
| HLE | RCE | Accuracy23.1 | 11 | 4mo ago | |
| GPQA | RCE | Accuracy48.9 | 11 | 4mo ago | |
| ARC-AGI 2 | RCE | Accuracy33.6 | 11 | 4mo ago | |
| CompA Attribute sub-task | CLAPScore | Text Attribute Accuracy44.28 | 11 | 4mo ago | |
| CompA Order sub-task | AQAScore | Text Score0.67 | 11 | 4mo ago | |
| NaturalBench | InternVL3.5-14B +FINER-Tuning | Accuracy35.5 | 10 | 4mo ago | |
| Cola | NegCLIP++ | Txt2Img Score33.33 | 10 | 4mo ago | |
| MAT-PREF OOD-property | Qwen3-8B + SFT + GRPO | Logical Question Accuracy60 | 8 | 1mo ago |