Scam Detection Benchmark Single-App Content
99.7AccuracyORACLE
Evaluation Results
| Method | Links | |
|---|---|---|
| ORACLEBackbone=Qwen3-4B, Category=ORACLE2026.05 | 99.7 | |
| FraudR1Category=Specialized Methods2026.05 | 98.9 | |
| SafeQAQCategory=Specialized Methods2026.05 | 98.4 | |
| ORACLEBackbone=Qwen3-1.7B, Category=ORACLE2026.05 | 98.1 | |
| ScamGPT-JCategory=Specialized Methods2026.05 | 97.8 | |
| AntiFraudCategory=Specialized Methods2026.05 | 97.5 | |
| Claude-4.1-OpusCategory=General LLMs2026.05 | 96.8 | |
| Claude-4-SonnetCategory=General LLMs2026.05 | 96.4 | |
| GPT-5.1Category=General LLMs2026.05 | 96.3 | |
| ORACLEBackbone=Qwen3-0.6B, Category=ORACLE2026.05 | 96.2 | |
| BotWarCategory=Specialized Methods2026.05 | 96.1 | |
| Gemini-3-FlashCategory=General LLMs2026.05 | 95.9 | |
| Kimi-2.5Category=General LLMs2026.05 | 95.7 | |
| GLM-4.6Category=General LLMs2026.05 | 95.6 | |
| Grok-4Category=General LLMs2026.05 | 94.8 | |
| DeepSeek-V4Category=General LLMs2026.05 | 94.1 |