Veracity Classification on ReMMDBench (500-sample split)
41.8AccuracyReMMD-Agent
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| ReMMD-AgentBackbone=GPT-5.22026.06 | 41.8 | 43.98 | 41.71 | 39.12 | |
| ReMMD-AgentBackbone=Qwen3.5-9B2026.06 | 37.2 | 39.72 | 37.11 | 37.18 | |
| ReMMD-AgentBackbone=Gemma4-31B2026.06 | 33.6 | 34.21 | 33.59 | 33.76 | |
| ManusBackbone=proprietary2026.06 | 33 | 33.58 | 33 | 33.13 | |
| ReMMD-AgentBackbone=Qwen3.6-27B2026.06 | 30.4 | 32.85 | 30.36 | 30.07 | |
| ChatGPTBackbone=proprietary2026.06 | 30.2 | 32.65 | 30.15 | 28.24 | |
| ReMMD-AgentBackbone=Qwen3.5-4B2026.06 | 29.2 | 32.07 | 29.12 | 28.58 | |
| T2-AgentBackbone=GPT-5.22026.06 | 28.2 | 29.91 | 28.14 | 26 | |
| MMD-AgentBackbone=GPT-5.22026.06 | 26.4 | 23.77 | 26.38 | 23.42 | |
| MMD-AgentBackbone=Qwen3.5-9B2026.06 | 26.2 | 26.16 | 26.24 | 26 | |
| T2-AgentBackbone=Qwen3.6-27B2026.06 | 26 | 24.07 | 26.13 | 23.4 | |
| MMD-AgentBackbone=Gemma4-31B2026.06 | 25.6 | 26.83 | 25.6 | 25.86 | |
| T2-AgentBackbone=Qwen3.5-9B2026.06 | 25.6 | 23.98 | 25.67 | 23.31 | |
| MMD-AgentBackbone=Qwen3.6-27B2026.06 | 25.4 | 26.3 | 25.44 | 25.36 | |
| MMD-AgentBackbone=Qwen3.5-4B2026.06 | 25 | 25.42 | 25.04 | 24.66 | |
| T2-AgentBackbone=Gemma4-31B2026.06 | 24.4 | 24.97 | 24.41 | 24.6 | |
| T2-AgentBackbone=Qwen3.5-4B2026.06 | 21.2 | 20.41 | 21.2 | 19.92 |