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: 分数 ↑.
| 排名 | 模型 | 公司 | 答案错误 次数 | 分数 | 测试正确 | 响应时间（平均） |
|---|---|---|---|---|---|---|
| #103 | DeepSeek V4 Pro high | DeepSeek | 4 | 6.0 | 8/21 | 65.2s |
| #102 | Gemma 4 26B A4B none | 10 | 6.0 | 8/21 | 5.91s | |
| #101 | Mimo V2 Omni none | Xiaomi | 10 | 6.0 | 8/21 | 2.44s |
| #100 | Grok Build 0.1 none | X AI | 7 | 6.0 | 7/19 | 28.7s |
| #99 | gpt-oss-120b medium | OpenAI | 9 | 6.1 | 9/21 | 22.3s |
| #98 | GLM 5 none | Z.ai | 12 | 6.1 | 9/21 | 4.03s |
| #97 | Gemini 2.5 Flash none | 12 | 6.2 | 9/21 | 875ms | |
| #96 | Ring-2.6-1T none | Inclusionai | 5 | 6.2 | 9/21 | 55.1s |
| #95 | Qwen3.5 Plus 2026-02-15 none | Qwen | 12 | 6.3 | 9/21 | 2.31s |
| #94 | GPT-5 Nano medium | OpenAI | 9 | 6.3 | 9/21 | 42.5s |
| #93 | Qwen3.6 Plus Preview medium | Qwen | 2 | 6.3 | 9/19 | 15.2s |
| #92 | Laguna M.1 medium | Poolside | 4 | 6.4 | 9/19 | 14.7s |
| #91 | GPT-5.5 none | OpenAI | 11 | 6.4 | 10/21 | 1.89s |
| #89 | Hy3 preview low | Tencent | 4 | 6.4 | 10/21 | 24.6s |
| #90 | Gemini 3.1 Flash Lite none | 11 | 6.4 | 9/21 | 1.06s |