AI BENCHY 失败分析
答案错误 失败
看看哪些 AI 模型最常遇到 答案错误，让你在选择前先发现稳定性风险。 排序方式: 失败次数 ↑.
| 排名 | 模型 | 公司 | 答案错误 次数 | 分数 | 测试正确 | 响应时间（平均） |
|---|---|---|---|---|---|---|
| #11 | Claude Opus 4.7 medium | Anthropic | 3 | 8.7 | 17/21 | 4.73s |
| #13 | Grok 4.20 Beta medium | X AI | 3 | 8.5 | 14/18 | 9.75s |
| #17 | GLM 5 medium | Z.ai | 3 | 8.3 | 15/21 | 33.5s |
| #20 | Gemini 3.5 Flash none | 3 | 8.1 | 15/21 | 9.93s | |
| #33 | Hy3 preview medium | Tencent | 3 | 7.7 | 14/21 | 16.3s |
| #35 | Gemini 3 PRO Preview medium | 3 | 7.6 | 14/21 | 9.05s | |
| #37 | Gemma 4 26B A4B medium | 3 | 7.6 | 14/21 | 63.4s | |
| #42 | GPT-5.2 medium | OpenAI | 3 | 7.5 | 13/21 | 16.9s |
| #43 | MiMo-V2.5-Pro medium | Xiaomi | 3 | 7.5 | 12/21 | 26.1s |
| #60 | Kimi K2.6 medium | Moonshot AI | 3 | 7.2 | 12/21 | 71.7s |
| #67 | MiniMax M3 medium | Minimax | 3 | 7.1 | 11/21 | 68.2s |
| #69 | Claude Opus 4.6 medium | Anthropic | 3 | 7.0 | 12/21 | 25.9s |
| #82 | Hy3 preview high | Tencent | 3 | 6.6 | 11/21 | 56.6s |
| #9 | GPT-5.5 medium | OpenAI | 4 | 8.8 | 17/21 | 38.0s |
| #15 | GPT-5.3-Codex medium | OpenAI | 4 | 8.4 | 15/21 | 16.2s |