反AI技巧：答案错误
反AI技巧
答案错误
看看哪些 AI 模型在 反AI技巧 上最容易遇到 答案错误，更快找出薄弱点。
140/140
筛选模型
没有模型匹配当前搜索和筛选条件。
| 排名 | 模型 | 公司 | 答案错误 次数 | 分类得分 | 总成本 | 测试正确 | 响应时间（平均） |
|---|---|---|---|---|---|---|---|
| #84 | Seed-2.0-Mini medium | Bytedance Seed | 1 | 6.6 | $0.101 | 2/4 | 74.7s |
| #85 | KAT-Coder-Pro V2.5 medium | Kwaipilot | 1 | 8.2 | $0.467 | 3/4 | 3.90s |
| #87 | GPT-5.6 Sol none | OpenAI | 1 | 8.3 | $0.524 | 3/4 | 1.27s |
| #93 | Gemini 3 Flash Preview none | 1 | 8.3 | $0.085 | 3/4 | 1.25s | |
| #97 | KAT-Coder-Pro V2.5 none | Kwaipilot | 1 | 8.7 | $0.476 | 3/4 | 4.58s |
| #98 | GLM 5V Turbo medium | Z.ai | 1 | 7.2 | $0.457 | 2/4 | 10.8s |
| #99 | Claude Opus 4.7 none | Anthropic | 1 | 8.3 | $0.505 | 3/4 | 2.12s |
| #101 | GLM 5.2 none | Z.ai | 1 | 8.3 | $0.128 | 3/4 | 3.70s |
| #105 | Qwen3.6 27B medium | Qwen | 1 | 8.3 | $0.779 | 3/4 | 12.6s |
| #110 | Gemini 3.1 Flash Lite Preview low | 1 | 8.3 | $0.646 | 3/4 | 2.12s | |
| #112 | Gemini 3.1 Flash Lite Preview none | 1 | 7.5 | $0.052 | 2/4 | 1.04s | |
| #118 | Claude Sonnet 5 none | Anthropic | 1 | 5.3 | $0.548 | 1/4 | 3.60s |
| #123 | GPT-5.6 Luna low | OpenAI | 1 | 8.3 | $0.249 | 3/4 | 2.21s |
| #126 | Gemini 3.1 Flash Lite minimal | 1 | 8.3 | $0.047 | 3/4 | 1.10s | |
| #127 | gpt-oss-120b medium | OpenAI | 1 | 6.7 | $0.019 | 2/4 | 10.2s |