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: 响应时间（平均） ↑.
| 排名 | 模型 | 公司 | 答案错误 次数 | 分数 | 测试正确 | 响应时间（平均） |
|---|---|---|---|---|---|---|
| #3 | Gemini 3.5 Flash low | 2 | 9.4 | 19/21 | 3.27s | |
| #74 | Qwen3.6 Max Preview none | Qwen | 10 | 6.9 | 11/21 | 3.30s |
| #117 | Qwen3.5-35B-A3B none | Qwen | 12 | 5.6 | 7/21 | 3.37s |
| #131 | Qwen3.5-122B-A10B none | Qwen | 13 | 5.3 | 6/21 | 3.41s |
| #68 | Claude Opus 4.8 none | Anthropic | 4 | 7.0 | 12/21 | 3.47s |
| #108 | Qwen3.5-Flash none | Qwen | 13 | 5.8 | 8/21 | 3.58s |
| #118 | Qwen3.6 27B none | Qwen | 11 | 5.6 | 7/21 | 3.72s |
| #153 | Qwen3.6 35B A3B none | Qwen | 13 | 4.6 | 4/21 | 3.73s |
| #40 | Gemini 3.1 Flash Lite Preview medium | 7 | 7.5 | 13/21 | 3.96s | |
| #98 | GLM 5 none | Z.ai | 12 | 6.1 | 9/21 | 4.03s |
| #85 | Gemma 4 31B none | 8 | 6.5 | 10/21 | 4.05s | |
| #112 | GLM 5.1 none | Z.ai | 13 | 5.7 | 7/21 | 4.10s |
| #114 | Qwen3.5 Plus 2026-04-20 none | Qwen | 12 | 5.7 | 7/21 | 4.39s |
| #116 | Hunter Alpha none | OpenRouter | 9 | 5.7 | 6/18 | 4.70s |
| #11 | Claude Opus 4.7 medium | Anthropic | 3 | 8.7 | 17/21 | 4.73s |