Artificial Analysis Coding Agents
按 Artificial Analysis 自己发布的样子呈现——他们的指标，他们的模型。点击模型名可前往其上游页面。
- 许可
- 未声明
- 更新频率
- 持续更新
- 抓取时间
- 2026-08-20
- 条目数
- 59 · 5 条未匹配
- Claude Opus 5
0.667 - GPT-5.6 Sol
0.666 - Fable 5 (max) (with fallback) 0.658
- Grok 4.5
0.644 - GPT-5.6 Terra 0.623
- GPT-5.5 0.615
- Kimi K3
0.613 - Claude Opus 4.8 0.606
- Muse Spark 1.2
0.605 - Qwen3.8 Max
0.598 - GPT-5.6 Luna 0.587
- Gemini 3.7 Flash
0.571
59 个有分数
| # | 模型 | ||||||
|---|---|---|---|---|---|---|---|
| 1 | Claude Opus 5Claude Code · xhighanthropic | 0.667 | 60.5% | 54.8% | 84.9% | $8.23 | 1418.81s |
| 2 | GPT-5.6 SolCodex · maxopenai | 0.666 | 68.7% | 43.3% | 87.7% | $7.08 | 610.08s |
| 3 | Fable 5 (max) (with fallback)Claude Code未匹配 | 0.658 | 66.1% | 48.9% | 82.5% | $11.70 | 1403.71s |
| 4 | Claude Opus 5Claude Code · maxanthropic | 0.655 | 63.1% | 48.9% | 84.5% | $8.95 | 1423.99s |
| 5 | GPT-5.6 SolCodex · xhighopenai | 0.651 | 67.0% | 42.2% | 86.1% | $5.24 | 443.62s |
| 6 | Grok 4.5Grok Build · highxai | 0.644 | 59.9% | 48.1% | 85.3% | $2.59 | 988.66s |
| 7 | GPT-5.6 SolCodex · highopenai | 0.641 | 64.9% | 44.9% | 82.5% | $4.14 | 379.29s |
| 8 | Claude Opus 5Claude Code · highanthropic | 0.634 | 60.8% | 49.2% | 80.2% | $3.80 | 802.32s |
| 9 | GPT-5.6 TerraCodex · maxopenai | 0.623 | 67.0% | 35.8% | 84.1% | $2.21 | 502.38s |
| 10 | Claude Opus 5Claude Code · mediumanthropic | 0.619 | 62.8% | 44.4% | 78.6% | $3.14 | 731.49s |
| 11 | GPT-5.5Codex · xhighopenai | 0.615 | 64.3% | 36.0% | 84.1% | $5.07 | 604.97s |
| 12 | Kimi K3Kimi Code CLIkimi | 0.613 | 63.7% | 36.6% | 83.7% | $3.18 | 1427.73s |
| 13 | Claude Opus 4.8Claude Code · maxanthropic | 0.606 | 55.8% | 46.8% | 79.4% | $7.70 | 1387.44s |
| 14 | GPT-5.6 SolCodex · mediumopenai | 0.606 | 64.0% | 40.1% | 77.8% | $2.99 | 310.18s |
| 15 | Muse Spark 1.2Muse Code · xhighmeta | 0.605 | 58.1% | 44.9% | 78.6% | $2.33 | 2519.42s |
| 16 | Qwen3.8 MaxClaude Codealibaba | 0.598 | 51.9% | 48.2% | 79.4% | $3.30 | 1791.91s |
| 17 | GPT-5.6 LunaCodex · maxopenai | 0.587 | 63.4% | 32.8% | 79.8% | $0.31 | 479.74s |
| 18 | Claude Opus 4.8Claude Code · xhighanthropic | 0.585 | 51.3% | 42.7% | 81.3% | $5.67 | 1062.37s |
| 19 | GPT-5.6 TerraCodex · xhighopenai | 0.571 | 58.4% | 32.3% | 80.6% | $1.52 | 412.95s |
| 20 | Gemini 3.7 FlashOpencode · highgoogle | 0.571 | 57.2% | 30.6% | 83.3% | $1.26 | 507.93s |
| 21 | Claude Opus 5Claude Code · lowanthropic | 0.568 | 56.9% | 39.2% | 74.2% | $2.18 | 569.29s |
| 22 | Muse Spark 1.2Opencode · xhighmeta | 0.568 | 52.8% | 44.1% | 73.4% | $1.91 | 1068.84s |
| 23 | Claude Opus 4.8Claude Code · highanthropic | 0.567 | 51.6% | 38.7% | 79.8% | $3.74 | 746.28s |
| 24 | Gemini 3.7 FlashAntigravity SDK v0.1.8 · highgoogle | 0.564 | 56.0% | 27.2% | 86.1% | $1.39 | 380.37s |
| 25 | GPT-5.6 TerraCodex · highopenai | 0.558 | 60.5% | 30.9% | 76.0% | $1.27 | 370.78s |
| 26 |  DeepSeek V4 FlashCodex · maxdeepseek | 0.555 | 42.8% | 38.7% | 84.9% | $0.07 | 877.44s |
| 27 | GPT-5.6 LunaCodex · xhighopenai | 0.547 | 56.6% | 31.2% | 76.2% | $0.25 | 395.38s |
| 28 | GPT-5.5Codex · mediumopenai | 0.544 | 56.6% | 30.6% | 75.8% | $2.75 | 384.13s |
| 29 | GPT-5.6 SolCodex · lowopenai | 0.536 | 53.4% | 34.4% | 73.0% | $1.72 | 221.85s |
| 30 | Claude Opus 4.8Claude Code · mediumanthropic | 0.536 | 49.3% | 36.0% | 75.4% | $3.26 | 742.54s |
| 31 | Muse Spark 1.1Opencode · xhighmeta | 0.535 | 54.3% | 33.3% | 73.0% | $1.43 | 755.21s |
| 32 | GPT-5.6 LunaCodex · highopenai | 0.514 | 53.4% | 29.0% | 71.8% | $0.19 | 339.05s |
| 33 | Claude Opus 4.7Claude Code · maxanthropic | 0.503 | 40.1% | 37.1% | 73.8% | $5.63 | 939.75s |
| 34 | Claude Opus 4.7Opencode · mediumanthropic | 0.503 | 39.5% | 36.0% | 75.4% | $2.93 | 731.68s |
| 35 | GPT-5.6 TerraCodex · mediumopenai | 0.478 | 45.7% | 28.2% | 69.4% | $0.72 | 256.39s |
| 36 | Claude Opus 4.8Claude Code · lowanthropic | 0.474 | 41.0% | 28.2% | 73.0% | $2.15 | 505.67s |
| 37 | Claude Opus 4.6Claude Code · mediumanthropic | 0.465 | — | 22.3% | 70.6% | $1.28 | 479.63s |
| 38 | GPT-5.5Cursor CLI · mediumopenai | 0.461 | 37.2% | 27.7% | 73.4% | $2.01 | 398.80s |
| 39 | Gemini 3.6 FlashOpencode · highgoogle | 0.456 | 41.3% | 21.8% | 73.8% | $2.08 | 625.26s |
| 40 | Claude Opus 4.7Cursor CLI · mediumanthropic | 0.454 | 31.6% | 33.9% | 70.6% | $2.68 | 816.85s |
| 41 | GPT-5.6 SolCodex · offopenai | 0.434 | 35.4% | 34.1% | 60.7% | $1.40 | 206.26s |
| 42 |  GLM-5.2Claude Codezai | 0.432 | 28.6% | 29.0% | 71.9% | $6.51 | 1505.09s |
| 43 | GPT-5.6 LunaCodex · mediumopenai | 0.424 | 36.6% | 27.2% | 63.5% | $0.09 | 202.33s |
| 44 | Claude Opus 4.7Claude Code · mediumanthropic | 0.405 | 27.4% | 22.6% | 71.4% | $1.68 | 379.61s |
| 45 | GPT-5.4Codex · mediumopenai | 0.391 | 25.0% | 22.3% | 69.8% | $2.42 | 425.07s |
| 46 | Composer 2.5Cursor CLI未匹配 | 0.382 | 15.9% | 31.2% | 67.5% | $0.08 | 572.81s |
| 47 | Composer 2.5 FastCursor CLI未匹配 | 0.382 | 15.9% | 31.2% | 67.5% | $0.55 | 406.08s |
| 48 | Claude Sonnet 4.6Claude Code · mediumanthropic | 0.376 | 28.9% | 19.6% | 64.3% | $2.01 | 807.42s |
| 49 | GPT-5.4Cursor CLI · mediumopenai | 0.371 | 16.7% | 28.2% | 66.3% | $1.55 | 485.68s |
| 50 | GPT-5.6 TerraCodex · lowopenai | 0.367 | 29.8% | 22.8% | 57.5% | $0.39 | 167.00s |
| 51 | GLM-5.1Claude Codezai | 0.361 | 18.6% | 24.7% | 65.1% | $4.33 | 1166.25s |
| 52 | Qwen3.7 PlusClaude Code · onalibaba | 0.360 | 19.2% | 23.7% | 65.1% | $6.23 | 633.82s |
| 53 | Kimi K2.6Claude Codekimi | 0.326 | 16.5% | 15.9% | 65.5% | $1.19 | 2460.81s |
| 54 | DeepSeek V4 ProClaude Code · highdeepseek | 0.314 | 8.6% | 19.9% | 65.9% | $0.27 | 1072.49s |
| 55 | Gemini 3.1 Pro (high)Gemini CLI · high未匹配 | 0.303 | 14.2% | 8.6% | 68.3% | $2.00 | 649.43s |
| 56 | Composer 2Cursor CLI未匹配 | 0.275 | 0.0% | 17.7% | 64.7% | $0.04 | 514.04s |
| 57 | GPT-5.6 LunaCodex · lowopenai | 0.251 | 10.3% | 15.3% | 49.6% | $0.04 | 115.41s |
| 58 | GPT-5.6 TerraCodex · offopenai | 0.237 | 13.3% | 18.5% | 39.3% | $0.30 | 107.57s |
| 59 | GPT-5.6 LunaCodex · offopenai | 0.204 | 6.5% | 17.5% | 37.3% | $0.07 | 149.67s |
各列衡量的是什么
- Artificial Analysis Coding Agent Index
- 越高越好 · 智能体编程
- DeepSWE
- 越高越好
- SWE-Atlas-QnA
- 越高越好
- Terminal-Bench v2
- 越高越好 · 智能体编程
- Cost
- 越低越好 · 成本
- Execution Time
- 越低越好