LOADING載入中
57 個模型
| 排名 | 模型 | 淨提升 | 成功率 | 讚揚 / 投訴 | 可引導性 | Bash 可靠性 | 工具幻覺 | 得分下界 | 得分上界 | 工作階段數 | 觀測數 | 每任務成本(P50) | 每任務輸出 Token(P50) | 分類 | 來源資料更新 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| #1 | Claude Opus 5.5 (Max)anthropic · Proprietary | 20.71%±3.20% | 42.34%±7.72% | 18.57%±2.02% | 15.00%±0.42% | 0.68%±0.17% | 17.9% | 22.01% | 14,358 | 2,477,322 | $6.31 | 120,135 | overall | 2026-11-08 | |
| #2 | Gemini 4 Carbon (High)google · Proprietary | 18.39%±1.15% | 39.54%±2.59% | 19.13%±0.65% | -5.26%±0.70% | 0.41%±0.20% | 15.92% | 17.34% | 53,721 | 8,594,005 | $1.54 | 50,900 | overall | 2026-11-08 | |
| #3 | Claude Sonnet 5.5 (Max)anthropic · Proprietary | 19.49%±4.45% | 31.56%±11.06% | 13.76%±2.81% | 12.70%±0.61% | 0.80%±0.13% | 13.15% | 18.84% | 8,670 | 1,241,679 | $3.42 | 149,861 | overall | 2026-11-08 | |
| #4 | Claude Fable 5.1 (Max)anthropic · Proprietary | 13.13%±3.64% | 29.82%±7.29% | 15.74%±1.86% | 10.02%±0.86% | 0.81%±0.12% | 12.36% | 16.32% | 17,468 | 2,440,710 | $6.87 | 56,226 | overall | 2026-11-08 | |
| #5 | GPT-6 Astra (Max)openai · Proprietary | 13.58%±3.49% | 30.87%±7.46% | 14.82%±1.83% | 7.78%±1.13% | 0.92%±0.08% | 12.21% | 16.27% | 15,520 | 1,318,529 | $7.04 | 43,779.75 | overall | 2026-11-08 | |
| #6 | GPT-6.1 Sol (Max)openai · Proprietary | 11.20%±2.03% | 28.42%±4.01% | 13.35%±1.07% | 4.98%±0.68% | 0.92%±0.08% | 11.41% | 13.64% | 30,514 | 3,003,038 | $0.98 | 35,186.5 | overall | 2026-11-08 | |
| #7 | GPT-6.1 Terra (Max)openai · Proprietary | 10.63%±2.62% | 20.45%±4.72% | 15.81%±1.45% | 13.31%±0.39% | 0.92%±0.08% | 10.75% | 13.46% | 23,276 | 7,778,687 | $0.8 | 43,743 | overall | 2026-11-08 | |
| #8 | Gemini 4 Argon (High)google · Proprietary | 14.07%±1.54% | 23.52%±3.06% | 13.53%±0.97% | -5.63%±1.13% | 0.43%±0.15% | 9.96% | 11.7% | 53,301 | 6,590,911 | $0.82 | 28,823 | overall | 2026-11-08 | |
| #9 | Muse Wonder Full (Max)meta · Proprietary | 11.02%±1.80% | 15.39%±3.02% | 10.43%±1.05% | 8.34%±0.44% | -0.12%±1.22% | 8.15% | 10.03% | 32,477 | 5,895,619 | — | 34,407 | overall | 2026-11-08 | |
| #10 | Claude Fable 5 (High)anthropic · Proprietary | 5.72%±2.23% | 14.60%±3.60% | 8.48%±1.38% | 10.37%±0.83% | 0.66%±0.19% | 6.57% | 8.83% | 51,053 | 3,506,790 | $3.42 | 22,703.5 | overall | 2026-11-08 | |
| #11 | Claude Opus 5 (Max)anthropic · Proprietary | 10.55%±2.49% | 12.46%±4.22% | 4.94%±1.61% | 9.84%±0.58% | 0.87%±0.09% | 6.2% | 8.79% | 31,928 | 4,646,044 | $4.14 | 57,561 | overall | 2026-11-08 | |
| #12 | GPT-6 Sol (Max)openai · Proprietary | 3.76%±3.40% | 11.20%±5.18% | 10.11%±1.70% | 11.04%±0.54% | 0.92%±0.08% | 5.45% | 8.56% | 12,836 | 3,219,770 | $1.76 | 40,170 | overall | 2026-11-08 | |
| #13 | Kimi K3.1 Pro (Max)moonshot · Kimi K3.1 Pro License | 2.84%±2.44% | 8.74%±3.55% | 7.45%±1.34% | 7.25%±0.53% | 0.92%±0.08% | 4.11% | 6.36% | 22,581 | 6,454,476 | $1.31 | 36,616.5 | overall | 2026-11-08 | |
| #14 | Hy4tencent · Proprietary | 8.06%±2.90% | 4.69%±4.74% | 5.01%±1.84% | 6.25%±0.78% | 0.08%±0.31% | 3.23% | 6.08% | 17,888 | 2,514,174 | $0.59 | 60,734.5 | overall | 2026-11-08 | |
| #15 | GPT-5.6 Sol (xHigh)openai · Proprietary | 1.87%±2.30% | 5.84%±3.40% | 6.09%±1.43% | 11.28%±0.67% | 0.92%±0.08% | 3.45% | 5.6% | 45,371 | 6,593,859 | $1.72 | 23,000.33 | overall | 2026-11-08 | |
| #16 | Grok 4.7 (xHigh)xai · Proprietary | 5.46%±3.11% | 2.79%±4.61% | -2.02%±2.05% | 9.55%±0.63% | 0.92%±0.08% | 1.18% | 4.12% | 18,848 | 3,715,248 | $2.91 | 79,491 | overall | 2026-11-08 | |
| #17 | Gemini 3.8 Flash (High)google · Proprietary | 0.95%±1.82% | 4.99%±2.54% | -0.64%±1.19% | 9.93%±0.40% | 0.46%±0.15% | 1.54% | 3.23% | 52,722 | 8,531,896 | $0.75 | 53,328 | overall | 2026-11-08 | |
| #18 | MiMo-V2.6-Proxiaomi · MIT | 4.93%±1.97% | 3.75%±2.78% | 0.80%±1.23% | 1.94%±0.65% | 0.04%±0.22% | 1.42% | 3.24% | 49,297 | 5,591,042 | $0.17 | 49,540.5 | overall | 2026-11-08 | |
| #19 | Muse Spark 1.3 (Max)meta · Proprietary | 3.67%±1.81% | 1.67%±2.57% | 2.94%±1.12% | 1.88%±0.80% | 0.74%±0.17% | 1.37% | 3.05% | 80,655 | 8,847,054 | $0.47 | 26,458.5 | overall | 2026-11-08 | |
| #20 | DeepSeek-V4.1-Flash-20261017 (Max)deepseek · MIT | 2.53%±1.27% | 1.65%±1.68% | 3.53%±0.70% | 2.63%±0.31% | 0.31%±0.11% | 1.53% | 2.62% | 78,504 | 22,460,118 | $0.15 | 85,928.5 | overall | 2026-11-08 | |
| #21 | Claude Sonnet 5 (High)anthropic · Proprietary | -0.53%±2.68% | -1.23%±3.54% | 3.02%±1.71% | 15.04%±0.54% | 0.50%±0.15% | 0.88% | 3.24% | 41,726 | 5,190,733 | $0.91 | 22,322 | overall | 2026-11-08 | |
| #22 | Muse Wonder (Max)meta · Proprietary | 5.41%±2.93% | -0.90%±3.94% | 2.60%±1.81% | 4.72%±0.97% | -0.81%±1.43% | 0.6% | 3.25% | 13,921 | 2,151,916 | $1.28 | 24,585 | overall | 2026-11-08 | |
| #23 | GPT-6 Luna (Max)openai · Proprietary | -0.56%±1.03% | -2.02%±1.40% | 3.76%±0.58% | 9.51%±0.28% | 0.92%±0.08% | 1.07% | 1.98% | 228,261 | 41,971,452 | $0.09 | 40,914 | overall | 2026-11-08 | |
| #24 | GLM-5.3 (Max)zai · MIT | 2.92%±1.79% | 4.37%±2.45% | -0.95%±1.14% | -5.60%±0.86% | 0.92%±0.08% | 0.18% | 1.8% | 95,597 | 11,344,109 | $1.09 | 65,001 | overall | 2026-11-08 | |
| #25 | Kimi K3 (Max)moonshot · Kimi K3 license | 3.47%±1.33% | 0.89%±1.76% | -0.36%±0.85% | -4.72%±0.65% | 0.92%±0.08% | -0.02% | 1.16% | 160,495 | 16,081,349 | $0.93 | 26,299 | overall | 2026-11-08 | |
| #26 | Hy4 previewtencent · Apache 2.0 | 4.16%±1.61% | -0.21%±2.20% | -0.38%±1.00% | -3.17%±0.50% | -0.01%±0.21% | -0.28% | 1.16% | 102,097 | 12,008,575 | $0.46 | 52,463.5 | overall | 2026-11-08 | |
| #27 | Claude Haiku 5.5 (High)anthropic · Proprietary | 1.34%±2.46% | -4.08%±3.11% | 2.66%±1.35% | 7.46%±0.55% | -1.73%±0.66% | -0.6% | 1.45% | 31,731 | 4,305,995 | $0.08 | 25,191 | overall | 2026-11-08 | |
| #28 | GLM-5.2 (Max)zai · MIT | 2.05%±1.73% | 1.96%±2.37% | -0.86%±1.19% | -4.50%±0.77% | 0.92%±0.08% | -0.39% | 1.2% | 98,965 | 9,275,648 | $0.9 | 56,399 | overall | 2026-11-08 | |
| #29 | Qwen3.8-Maxalibaba · Proprietary | 4.66%±1.45% | 0.01%±2.04% | -0.74%±0.98% | -3.27%±0.68% | -0.84%±0.22% | -0.34% | 0.99% | 80,457 | 8,481,806 | $1.24 | 51,276.25 | overall | 2026-11-08 | |
| #30 | MiMo-V2.6-Flashxiaomi · MIT | 4.99%±2.22% | -0.69%±3.07% | 2.31%±1.36% | -4.38%±0.99% | -4.00%±1.95% | -1% | 1.18% | 32,025 | 3,803,116 | $0.07 | 45,540.5 | overall | 2026-11-08 | |
| #31 | Inkling 2 Smallthinky · Proprietary | 2.11%±5.90% | -3.57%±7.84% | 2.08%±3.55% | 2.57%±1.85% | -1.53%±1.28% | -2.48% | 2.65% | 4,063 | 529,043 | $0.26 | 49,607 | overall | 2026-11-08 | |
| #32 | Grok 4.6 (xHigh)xai · Proprietary | -2.50%±2.78% | -3.21%±3.42% | -1.00%±1.73% | 10.18%±1.01% | 0.92%±0.08% | -1.36% | 1.05% | 30,247 | 3,795,827 | $1.46 | 34,481 | overall | 2026-11-08 | |
| #33 | GPT-5.6 Terra (xHigh)openai · Proprietary | -7.48%±1.83% | -2.42%±2.35% | 0.17%±1.11% | 0.64%±0.87% | 0.91%±0.08% | -2.68% | -1.09% | 44,425 | 3,806,197 | $0.64 | 17,192.5 | overall | 2026-11-08 | |
| #34 | Gemini 3.7 Flash (High)google · Proprietary | -3.47%±1.98% | -2.49%±2.44% | -6.36%±1.43% | 6.30%±0.86% | 0.64%±0.14% | -2.79% | -1% | 63,336 | 6,197,412 | $0.52 | 43,088.5 | overall | 2026-11-08 | |
| #35 | GPT-6 Luna (Medium)openai · Proprietary | -6.32%±1.22% | -6.98%±1.45% | 0.14%±0.68% | 6.82%±0.42% | 0.92%±0.08% | -2.46% | -1.47% | 76,341 | 10,316,806 | $0.03 | 9,628 | overall | 2026-11-08 | |
| #36 | MiniMax-M3.1-Flashminimax · MiniMax Community License | 0.35%±3.70% | -6.89%±4.40% | -3.47%±2.20% | -5.26%±0.92% | 0.23%±0.22% | -4.26% | -1.26% | 13,089 | 2,724,454 | $0.24 | 74,413 | overall | 2026-11-08 | |
| #37 | GLM-5.3-Flashzai · MIT | 1.68%±1.66% | -4.62%±1.94% | -4.73%±1.11% | -11.73%±0.89% | 0.92%±0.08% | -3.51% | -2.09% | 95,526 | 13,665,236 | $0.17 | 50,249.75 | overall | 2026-11-08 | |
| #38 | Step 5 Previewstepfun · Proprietary | -0.66%±3.23% | -6.27%±4.09% | -1.07%±1.91% | -11.67%±1.38% | -1.58%±0.51% | -4.83% | -2.02% | 17,051 | 2,852,835 | $0.66 | 47,706.5 | overall | 2026-11-08 | |
| #39 | Gemini 3.1 Pro Previewgoogle · Proprietary | -11.54%±2.88% | -1.03%±3.92% | -7.65%±2.75% | 1.99%±3.29% | -0.19%±0.75% | -5.78% | -2.85% | 87,553 | 2,960,870 | $0.55 | 13,353 | overall | 2026-11-08 | |
| #40 | Qwen3.8-27Balibaba · Apache 2.0 | -2.92%±2.01% | -6.57%±2.46% | -5.38%±1.40% | -11.90%±1.24% | 0.10%±0.19% | -5.47% | -3.74% | 63,984 | 8,412,983 | $0.4 | 54,184 | overall | 2026-11-08 | |
| #41 | Qwen3.8-Flash-Nextalibaba · qwen-community-1.0 | 0.00%±1.80% | -5.95%±2.14% | -2.89%±1.11% | -22.28%±0.79% | -1.01%±0.19% | -5.42% | -3.91% | 109,992 | 17,026,555 | $0.23 | 62,354 | overall | 2026-11-08 | |
| #42 | Mistral Large 4mistral · Modified MIT | -5.32%±4.76% | -12.15%±5.17% | -6.93%±3.27% | -5.90%±1.99% | -0.03%±0.93% | -8.02% | -4.15% | 5,559 | 519,381 | $0.32 | 25,933 | overall | 2026-11-08 | |
| #43 | GPT-6 Luna (Low)openai · Proprietary | -15.64%±1.83% | -11.02%±1.89% | -3.26%±0.97% | 1.97%±0.69% | 0.92%±0.08% | -6.92% | -5.53% | 43,861 | 3,615,936 | $0.01 | 3,796 | overall | 2026-11-08 | |
| #44 | Beam 1.0 501B A23Breflection · Apache 2.0 | -11.26%±6.22% | -8.10%±6.95% | -9.47%±3.68% | 3.51%±2.79% | -1.79%±1.50% | -8.97% | -3.86% | 12,257 | 695,650 | — | 13,873 | overall | 2026-11-08 | |
| #45 | Qwen3.7-Maxalibaba · Proprietary | -10.69%±2.86% | -10.29%±3.08% | -11.77%±2.26% | 2.66%±2.46% | -0.63%±0.62% | -8.36% | -5.89% | 43,146 | 2,297,138 | $0.35 | 15,879.5 | overall | 2026-11-08 | |
| #46 | Hy3tencent · Apache 2.0 | -12.09%±3.22% | -8.60%±3.91% | -2.59%±1.97% | -13.02%±3.51% | -4.74%±2.25% | -9.14% | -6.21% | 32,686 | 1,233,941 | $0.08 | 31,100 | overall | 2026-11-08 | |
| #47 | Muse Spark 1.1meta · Proprietary | -6.53%±1.55% | -13.78%±1.53% | -14.64%±1.32% | -2.58%±1.48% | 0.88%±0.09% | -8.52% | -7.19% | 149,186 | 8,493,252 | — | — | overall | 2026-11-08 | |
| #48 | Qwen3.7-Plusalibaba · Proprietary | -10.36%±3.73% | -12.01%±3.89% | -11.33%±2.65% | 0.45%±2.84% | -1.95%±1.20% | -9.45% | -6.29% | 24,539 | 1,286,452 | $0.13 | 16,207 | overall | 2026-11-08 | |
| #49 | Muse Spark 1.2 (xHigh)meta · Proprietary | -9.30%±2.19% | -13.80%±2.11% | -11.81%±1.50% | -3.87%±1.12% | 0.73%±0.21% | -8.88% | -7.17% | 54,770 | 3,866,021 | $0.41 | 25,811 | overall | 2026-11-08 | |
| #50 | Spark-X2.6synlan · Proprietary | -9.03%±3.43% | -18.79%±2.96% | -12.44%±2.00% | -8.36%±1.83% | -2.09%±1.09% | -11.59% | -9.08% | 11,434 | 1,265,738 | $0.09 | 36,428.5 | overall | 2026-11-08 | |
| #51 | MiniMax-M3minimax · MiniMax Community License | -17.68%±2.92% | -16.33%±2.67% | -10.34%±2.02% | -2.84%±1.74% | -1.13%±1.96% | -11.61% | -9.24% | 44,355 | 3,705,686 | $0.32 | 24,762 | overall | 2026-11-08 | |
| #52 | Mistral Medium 3.5mistral · Modified MIT | -17.71%±4.06% | -18.72%±3.74% | -16.65%±3.00% | -7.06%±3.82% | -6.08%±3.19% | -15.73% | -12.13% | 13,935 | 675,056 | $0.4 | 15,136.5 | overall | 2026-11-08 | |
| #53 | Nemotron 3 Ultranvidia · OpenMDW-1.1 | -22.23%±7.71% | -17.62%±7.73% | -15.45%±5.85% | -15.93%±11.35% | -1.42%±2.18% | -17.67% | -11.08% | 13,691 | 232,397 | $0.12 | 7,949.5 | overall | 2026-11-08 | |
| #54 | NVIDIA-Nemotron-3.5-Super-VL-120B-A12B (NVFP4)nvidia · OpenMDW-1.1 | -18.95%±2.10% | -23.66%±1.61% | -17.75%±1.28% | -11.66%±1.35% | -6.96%±1.21% | -17.04% | -15.47% | 29,893 | 2,205,171 | — | 14,532.25 | overall | 2026-11-08 | |
| #55 | Inklingthinky · Apache 2.0 | -26.34%±3.96% | -25.48%±2.53% | -21.72%±3.37% | 0.24%±3.13% | -3.41%±2.47% | -18.63% | -15.52% | 44,032 | 1,739,325 | $0.34 | 15,282 | overall | 2026-11-08 | |
| #56 | Inkling-Smallthinky · Apache 2.0 | -27.30%±4.93% | -22.43%±3.86% | -26.70%±3.86% | 0.97%±3.38% | -3.62%±2.36% | -19.71% | -15.65% | 14,747 | 568,000 | $0.18 | 20,902 | overall | 2026-11-08 | |
| #57 | Solar Pro 4upstage · Proprietary | -31.29%±5.90% | -24.73%±3.92% | -20.10%±4.07% | -45.97%±9.26% | -3.66%±1.69% | -25.02% | -20.65% | 11,012 | 559,330 | $0.04 | 11,348 | overall | 2026-11-08 |
用113個從零編寫的長程真實軟體工程任務評測程式設計 Agent,同時呈現準確率(Pass@1)與實際花費/步數的性價比曲線,避免題目訓練污染。
在新分頁開啟獨立跑分,不依賴廠商自報資料;綜合 Intelligence Index 涵蓋推理/程式設計/科學/知識,並結合速度、價格、上下文視窗等工程指標。
在新分頁開啟每月更新全新題目防污染,全自動客觀評分(唯一正確答案),聚焦數學/程式碼/推理/資料分析,為學術圈公認的權威。
在新分頁開啟中文原創題庫,深度貼合中文語境與政策文化,涵蓋數學推理、程式碼、幻覺控制、智慧代理規劃六大能力。
在新分頁開啟上海 AI 實驗室出品,20餘家企業共建,五大能力面向、70+ 資料集,中英文模型均涵蓋,並包含真人投票的 Compass Arena。
在新分頁開啟本頁對戰資料的官方原始來源,累計投票超過700萬次,涵蓋370+模型,逐票更新,介面本身也值得參考。
在新分頁開啟