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 ↑.
209/209
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） |
|---|---|---|---|---|---|---|---|
| #133 | Gemini 3 PRO Preview medium | 3 | 6.0 | $0.385 | 14/21 | 9.05s | |
| #135 | Hy3 preview high | Tencent | 3 | 5.9 | $0.048 | 11/21 | 56.6s |
| #3 | GPT-5.6 Sol low | OpenAI | 4 | 9.5 | $0.971 | 18/22 | 8.79s |
| #4 | GPT-5.6 Sol medium | OpenAI | 4 | 9.4 | $1.316 | 18/22 | 11.4s |
| #5 | GPT-5.6 Sol high | OpenAI | 4 | 9.4 | $1.234 | 18/22 | 11.7s |
| #10 | GPT-5.5 medium | OpenAI | 4 | 9.0 | $4.137 | 18/22 | 38.4s |
| #13 | GPT-5.3-Codex medium | OpenAI | 4 | 8.9 | $0.920 | 16/22 | 17.0s |
| #16 | Muse Spark 1.1 medium | Meta | 4 | 8.6 | $1.357 | 15/22 | 25.0s |
| #23 | Claude Sonnet 5 medium | Anthropic | 4 | 8.3 | $0.922 | 16/22 | 12.5s |
| #27 | Muse Spark 1.1 high | Meta | 4 | 8.1 | $1.694 | 12/22 | 31.5s |
| #28 | Inkling high | Thinkingmachines | 4 | 8.0 | $1.006 | 15/22 | 64.2s |
| #32 | Inkling medium | Thinkingmachines | 4 | 8.0 | $0.391 | 15/22 | 16.2s |
| #40 | Claude Sonnet 4.6 medium | Anthropic | 4 | 7.8 | $2.057 | 14/22 | 25.9s |
| #41 | Claude Opus 4.8 low | Anthropic | 4 | 7.8 | $2.077 | 16/22 | 12.7s |
| #49 | GLM 5 Turbo medium | Z.ai | 4 | 7.6 | $0.323 | 14/21 | 23.0s |