| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| BBH | DECENTMEM | Accuracy90.5 | 99 | 1mo ago | |
| SCALE MultiChallenge | BRAID | Accuracy65.1 | 81 | 1mo ago | |
| LLaVA Bench (val) | MARS | Perplexity2.1875 | 44 | 4mo ago | |
| BBH (val) | G2IS | Accuracy65.81 | 42 | 4mo ago | |
| Video-TT | OmniJigsaw (CMM) | Accuracy46.5 | 39 | 1mo ago | |
| SciFact (test) | LLM annotation | Macro-F176.17 | 37 | 1mo ago | |
| VitaminC (test) | EvoPool | Macro-F178.4 | 37 | 1mo ago | |
| FEVER (test) | EvoPool | Macro F185.18 | 37 | 1mo ago | |
| BBH | l2-guided Selection | Accuracy (%)89.14 | 28 | 1mo ago | |
| GraphRAG-Bench | HippoRAG 2 | Rel86.02 | 27 | 25d ago | |
| vcrbench | VideoLatent-7B (Ours) | Accuracy53.1 | 24 | 1mo ago | |
| GAIA Text | Accuracy76.4 | 19 | 3mo ago | ||
| BBH | Acc83.03 | 16 | 3mo ago | ||
| LLaVA-Bench Wilder | Qwen3-VL-4B (Base) | Score74.3 | 14 | 1mo ago | |
| Frames | Tongyi DeepResearch 30B | Accuracy90.6 | 13 | 3mo ago | |
| Humanity's Last Exam (HLE) | gemini-2.5-pro | Pass@1 Score18.4 | 13 | 4mo ago | |
| AIME | PRR | Decoding Speedup1.64 | 12 | 25d ago | |
| Arena-Hard 2.0 (test) | TCR | Overall Accuracy52.9 | 12 | 2mo ago | |
| AQuA | Accuracy28.35 | 12 | 2mo ago | ||
| GraphRAG-Bench Medical Dataset | HyperSU | Accuracy (ACC)75.03 | 9 | 25d ago | |
| TOMATO | Qwen3-VL-8B + SynRL | Accuracy38.1 | 9 | 4mo ago | |
| GSM8K (test) | Full Repetition | Accuracy87.3 | 8 | 22d ago | |
| MMLU-Pro (test) | Accuracy28.6 | 8 | 22d ago | ||
| Retrieval Quality Evaluation (test) | FlowRAG | Recall91.89 | 8 | 1mo ago | |
| G-Bench Medical (val) | MemGraphRAG | Recall90.42 | 8 | 1mo ago |
