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: 总成本 ↑.
215/215
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 | 分数 | 总成本 | 测试正确 | 响应时间（平均） |
|---|---|---|---|---|---|---|---|
| #136 | Step 3.5 Flash medium | Stepfun | 4 | 6.0 | $0.108 | 11/21 | 174.2s |
| #134 | GPT-5 Nano medium | OpenAI | 9 | 6.1 | $0.114 | 9/22 | 54.9s |
| #68 | Gemini 3.1 Flash Lite Preview medium | 7 | 7.3 | $0.115 | 13/22 | 4.61s | |
| #69 | Gemini 3.1 Flash Lite medium | 7 | 7.3 | $0.117 | 13/22 | 4.27s | |
| #132 | Qwen3.5 Plus 2026-04-20 none | Qwen | 12 | 6.1 | $0.122 | 8/22 | 13.6s |
| #161 | Kimi K2.5 none | Moonshot AI | 15 | 5.5 | $0.127 | 6/22 | 19.2s |
| #101 | GLM 5.2 none | Z.ai | 8 | 6.6 | $0.128 | 12/22 | 9.34s |
| #57 | GPT-5.4 Nano medium | OpenAI | 8 | 7.5 | $0.138 | 12/22 | 13.2s |
| #120 | Qwen3.5-Flash medium | Qwen | 4 | 6.2 | $0.139 | 12/22 | 84.8s |
| #165 | GPT-5.6 Luna none | OpenAI | 14 | 5.4 | $0.142 | 6/22 | 1.50s |
| #95 | Gemini 3.5 Flash-Lite low | 9 | 6.7 | $0.145 | 12/22 | 2.25s | |
| #170 | Inkling none | Thinkingmachines | 13 | 5.2 | $0.147 | 6/22 | 3.50s |
| #178 | MiniMax M2.7 medium | Minimax | 6 | 5.0 | $0.163 | 5/22 | 41.3s |
| #157 | GLM 5.1 none | Z.ai | 13 | 5.5 | $0.164 | 7/22 | 6.70s |
| #200 | GLM 4.7 Flash medium | Z.ai | 9 | 4.3 | $0.166 | 4/22 | 142.6s |