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） ↓.
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） |
|---|---|---|---|---|---|---|---|
| #161 | Kimi K2.5 none | Moonshot AI | 15 | 5.5 | $0.127 | 6/22 | 19.2s |
| #48 | GPT-5.6 Luna high | OpenAI | 7 | 7.7 | $1.017 | 15/22 | 18.7s |
| #179 | DeepSeek V3.2 none | DeepSeek | 7 | 5.0 | $0.054 | 6/22 | 18.3s |
| #20 | Claude Fable 5 medium | Anthropic | 2 | 8.6 | $3.478 | 17/22 | 17.2s |
| #213 | Nemotron 3 Nano Omni 30b A3b Reasoning medium | NVIDIA | 7 | 3.4 | $0.000 | 4/19 | 17.1s |
| #16 | GPT-5.3-Codex medium | OpenAI | 4 | 8.9 | $0.920 | 16/22 | 17.0s |
| #110 | Gemini 3.1 Flash Lite Preview low | 7 | 6.5 | $0.646 | 13/22 | 16.7s | |
| #106 | Hy3 preview medium | Tencent | 3 | 6.5 | $0.018 | 14/21 | 16.3s |
| #111 | Gemini 3.1 Flash Lite low | 9 | 6.5 | $0.621 | 12/22 | 16.3s | |
| #36 | Inkling medium | Thinkingmachines | 4 | 8.0 | $0.391 | 15/22 | 16.2s |
| #150 | KAT-Coder-Air V2.5 high | Kwaipilot | 9 | 5.6 | $0.077 | 7/22 | 15.9s |
| #23 | Grok 4.5 low | X AI | 6 | 8.4 | $0.935 | 16/22 | 15.6s |
| #181 | Qwen3.6 Plus Preview medium | Qwen | 2 | 4.9 | $0.000 | 9/19 | 15.2s |
| #4 | Gemini 3.5 Flash high | 1 | 9.5 | $1.976 | 20/22 | 15.1s | |
| #2 | Gemini 3.6 Flash high | 1 | 9.7 | $1.785 | 21/22 | 14.9s |