Machine-Generated Text Detection on MAGE
99.1AUROC (Avg)MELD
Evaluation Results
| Method | Links | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| MELDEvaluation Protocol=Supervised2026.05 | 99.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Human-as-OutlierEvaluation Protocol=Supervised2026.05 | 98 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ModernBERT-DetectEvaluation Protocol=Supervised2026.05 | 97.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RoBERTa-Large (OpenAI)Evaluation Protocol=Supervised2026.05 | 77.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| TSD+Category=Ours2026.01 | 75.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DDCategory=Ours2026.01 | 71.72 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| TSDCategory=Ours2026.01 | 71.56 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LVCategory=Ours2026.01 | 71.26 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LastdeCategory=Temporal2026.01 | 70.71 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Fast-DetectCategory=Global2026.01 | 69.69 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DiveyeCategory=Temporal2026.01 | 69.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GLTREvaluation Protocol=Zero-shot2026.05 | 64.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RADAREvaluation Protocol=Supervised2026.05 | 61.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| BinocularsEvaluation Protocol=Zero-shot2026.05 | 60.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FourierGPTCategory=Temporal2026.01 | 60.34 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RankCategory=Global2026.01 | 58.99 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Fast-DetectGPTEvaluation Protocol=Zero-shot2026.05 | 57.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LikelihoodCategory=Global2026.01 | 56.81 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RoBERTa-ChatGPTEvaluation Protocol=Supervised2026.05 | 56.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Log-RankCategory=Global2026.01 | 55.91 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| UCECategory=Temporal2026.01 | 54.17 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| EntropyCategory=Global2026.01 | 51.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LLRCategory=Global2026.01 | 51.78 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RepreGuardEvaluation Protocol=Supervised2026.05 | 47.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| BinocularsAlgorithm=vanilla2025.05 | — | 85.12 | 83.01 | 84 | 86.62 | 82.56 | 88.04 | 74.6 | 84.5 | 56.04 | 71.37 | 28.52 | 44.2 | — | |
| BinocularsAlgorithm=MCP2025.05 | — | 85.12 | 83.63 | 84.04 | 86.74 | 82.68 | 88.07 | 77.36 | 86.32 | 75.8 | 85.77 | 73.32 | 84.49 | — | |
| BinocularsAlgorithm=vanilla2025.05 | — | 85.12 | 83.01 | 84 | 86.62 | 82.56 | 88.04 | 74.6 | 84.5 | 56.04 | 71.37 | 28.52 | 44.2 | — | |
| BinocularsAlgorithm=MCP2025.05 | — | 84.8 | 82.65 | 83.44 | 86.18 | 82.52 | 88.02 | 77.36 | 86.32 | 75.8 | 85.77 | 73.32 | 84.49 | — | |
| DetectGPTAlgorithm=vanilla2025.05 | — | 59.36 | 66.21 | 38.2 | 51.54 | 19.8 | 31.73 | 7.64 | 13.94 | 3.88 | 7.4 | 2.04 | 3.98 | — | |
| DetectGPTAlgorithm=MCP2025.05 | — | 58.16 | 65.6 | 39.92 | 53.11 | 24.76 | 38.01 | 11.4 | 20.07 | 6.36 | 11.81 | 2.92 | 5.64 | — | |
| DetectRouterEncoder=LLM2Vec-LLaMA32026.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | 77.92 | |
| DiveyeSurrogate Model=LLaMA3-8B2026.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | 69.6 | |
| EntropyAlgorithm=vanilla2025.05 | — | 58.96 | 65.88 | 44.76 | 57.84 | 31.76 | 46.43 | 22.68 | 36.38 | 15.44 | 26.52 | 10.52 | 18.95 | — | |
| EntropyAlgorithm=MCP2025.05 | — | 58.76 | 66.47 | 43.2 | 56.63 | 31.72 | 46.46 | 21.44 | 34.85 | 16.12 | 27.55 | 10.44 | 18.84 | — | |
| EntropySurrogate Model=LLaMA3-8B2026.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | 51.9 | |
| Fast-DetectGPTAlgorithm=vanilla2025.05 | — | 80.36 | 80.15 | 77.08 | 82.4 | 72.92 | 81.99 | 66.04 | 78.6 | 57 | 72.15 | 43.08 | 60 | — | |
| Fast-DetectGPTAlgorithm=MCP2025.05 | — | 82.12 | 81.57 | 79.72 | 84.08 | 77.28 | 84.83 | 72.44 | 83.02 | 67.92 | 80.4 | 61.24 | 75.64 | — | |
| Fast-DetectGPTAlgorithm=vanilla2025.05 | — | 80.36 | 80.15 | 77.08 | 82.4 | 72.92 | 81.99 | 66.04 | 78.6 | 57 | 72.15 | 43.08 | 60 | — | |
| Fast-DetectGPTAlgorithm=MCP2025.05 | — | 82.12 | 81.57 | 79.72 | 84.08 | 77.28 | 84.83 | 72.44 | 83.02 | 65.92 | 78.91 | 51.4 | 67.61 | — | |
| Fast-DetectGPTSurrogate Model=LLaMA3-8B2026.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | 69.69 | |
| FourierGPTSurrogate Model=LLaMA3-8B2026.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | 60.34 | |
| LastdeSurrogate Model=LLaMA3-8B2026.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | 70.71 | |
| LikelihoodAlgorithm=vanilla2025.05 | — | 78.52 | 79.12 | 73.6 | 80.14 | 68.88 | 79.23 | 61.56 | 75.28 | 49.84 | 66.08 | 37.2 | 54.02 | — | |
| LikelihoodAlgorithm=MCP2025.05 | — | 79.56 | 80.14 | 74.6 | 80.81 | 69.2 | 79.5 | 62.96 | 76.26 | 53.08 | 68.88 | 35.44 | 52.21 | — | |
| LikelihoodSurrogate Model=LLaMA3-8B2026.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | 56.81 | |
| LLRSurrogate Model=LLaMA3-8B2026.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | 51.78 | |
| Log-RankAlgorithm=vanilla2025.05 | — | 78.64 | 79.18 | 74.12 | 80.43 | 69.48 | 79.62 | 61.64 | 75.34 | 51 | 67.11 | 45.52 | 62.34 | — | |
| Log-RankAlgorithm=MCP2025.05 | — | 79.28 | 80.11 | 74.52 | 80.95 | 70.24 | 80.24 | 64.24 | 77.17 | 55 | 70.51 | 44.6 | 61.53 | — | |
| LogRankSurrogate Model=LLaMA3-8B2026.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | 55.91 | |
| OpenAI-DAlgorithm=vanilla2025.05 | — | 82.84 | 81.68 | 71.72 | 78.92 | 63.08 | 75.06 | 54.96 | 70.03 | 49.96 | 66.19 | 43.08 | 60.02 | — | |
| OpenAI-DAlgorithm=MCP2025.05 | — | 80.96 | 80.96 | 74.04 | 80.74 | 67.68 | 78.5 | 59.36 | 73.68 | 53.16 | 69.02 | 44.76 | 61.7 | — | |
| RankSurrogate Model=LLaMA3-8B2026.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | 58.99 | |
| SupervisedRepresentations=LLaMA3-8B2026.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | 73.92 | |
| UCESurrogate Model=LLaMA3-8B2026.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | 54.17 |