编程 排名
看看哪些 AI 模型在 编程 上表现最好，哪些更稳定，以及差距主要出现在哪里。 排序方式: 测试正确 ↑.
408/408
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 | 编程 得分 | 分数 | 总成本 | 测试正确 | 响应时间（平均） |
|---|---|---|---|---|---|---|---|
| #64 | Claude Haiku 5.5 low | Anthropic | 0.8 | 0.9 | $0.061 | 2/3 | 7.30s |
| #67 | Muse Spark 1.3 low | Meta | 0.8 | 0.9 | $0.911 | 2/3 | 20.9s |
| #68 | Claude Sonnet 5 medium | Anthropic | 0.9 | 0.9 | $1.438 | 2/3 | 17.3s |
| #71 | Qwen3.8 2.4T A95B low | Qwen | 0.8 | 0.8 | $4.016 | 2/3 | 172.5s |
| #74 | Gemini 3.1 Pro Preview medium | 0.8 | 0.8 | $1.585 | 2/3 | 40.2s | |
| #75 | Inkling high | Thinkingmachines | 0.9 | 0.8 | $1.238 | 2/3 | 148.4s |
| #79 | GPT-5.6 Terra high | OpenAI | 0.8 | 0.8 | $0.698 | 2/3 | 9.14s |
| #88 | GLM 5.2 medium | Z.ai | 0.8 | 0.8 | $1.007 | 2/3 | 41.0s |
| #90 | Seed-2.0-Lite medium | Bytedance Seed | 0.8 | 0.8 | $0.271 | 2/3 | 156.7s |
| #93 | Claude Opus 4.7 medium | Anthropic | 0.8 | 0.8 | $3.636 | 2/3 | 13.0s |
| #94 | DeepSeek V4 Flash 0423 high | DeepSeek | 0.8 | 0.8 | $0.261 | 2/3 | 50.6s |
| #97 | Nemotron 3 Ultra medium | NVIDIA | 0.8 | 0.8 | $0.674 | 2/3 | 26.5s |
| #99 | Gemini 3.8 Flash medium | 0.8 | 0.8 | $0.769 | 2/3 | 11.7s | |
| #103 | GLM 5.3 FlashX max | Z.ai | 0.8 | 0.8 | $0.623 | 2/3 | 24.6s |
| #104 | Claude Fable 5.1 high | Anthropic | 0.8 | 0.8 | $3.471 | 2/3 | 13.9s |