OnlySOTA

Artificial Analysis LLM Leaderboard

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。

发布方
Artificial Analysis
许可
未声明
更新频率
持续更新
抓取时间
2026-08-22
条目数
271 · 1 条未匹配
按 Artificial Analysis Intelligence Index 排序的前 12 名,每个模型一行
  1. Claude Opus 5 63.1
  2. Claude Fable 5 (Opus 4.8 fallback) 62.1
  3. GPT-5.6 Sol 60.9
  4. Grok 4.6 60.9
  5. Kimi K3 59.7
  6. GLM-5.3 59.5
  7. Qwen3.8 Max 58.1
  8. Qwen3.8 2.4T A95B 57.7
  9. Muse Spark 1.2 56.8
  10. GPT-5.6 Terra 56.6
  11. Gemini 3.7 Flash 56.0
  12. Grok 4.5 55.8

263 个有分数 · 8 个未跑

#模型
1Claude Opus 5adaptive · maxanthropic63.193.2%54.9%55.7%29.1%42.1%89.1%67.2%75.7%37.178.059.239.2%84.7%$10.0059/s42.96s
2Claude Opus 5adaptive · xhighanthropic62.593.7%54.4%55.0%27.7%43.3%88.0%65.7%76.3%35.477.058.440.5%84.0%$10.0059/s20.79s
3Claude Fable 5 (Opus 4.8 fallback)anthropic62.192.6%55.5%60.2%28.6%38.1%84.6%61.9%76.7%43.376.556.636.4%63.5%$20.0071/s118.65s
4Claude Opus 5adaptive · highanthropic61.593.7%52.8%54.3%28.3%44.7%87.6%61.6%76.3%33.776.556.138.8%82.4%$10.0059/s15.09s
5GPT-5.6 Solmaxopenai60.994.1%49.5%56.1%32.3%44.3%88.0%61.1%77.7%22.077.457.87.8%72.7%83.4%$11.2571/s100.97s
6Grok 4.6highxai60.994.9%42.9%53.6%17.1%50.7%88.4%62.4%75.0%30.576.858.765.7%$3.0069/s52.02s
7Grok 4.6xhighxai60.093.5%44.1%51.6%19.7%43.3%88.0%63.3%75.7%29.375.956.676.0%$3.0064/s59.23s
8Kimi K3maxkimi59.793.5%46.9%58.7%23.4%46.0%85.0%58.9%82.7%19.776.254.346.8%80.5%$6.0038/s2.86s
9GLM-5.3maxzai59.591.7%42.3%56.5%19.1%50.3%83.9%63.4%76.3%14.374.859.170.4%$2.1593/s1.92s
10GPT-5.6 Solxhighopenai59.093.1%47.3%56.0%28.6%38.1%89.5%58.9%76.3%21.078.353.68.1%71.0%82.7%$11.2573/s26.79s
11Grok 4.6mediumxai59.093.5%42.1%54.6%17.7%44.3%84.3%62.1%72.7%28.074.456.376.0%$3.0067/s27.45s
12Claude Opus 5adaptive · mediumanthropic58.691.9%51.3%50.7%26.9%38.6%86.1%56.0%78.7%31.074.350.439.3%81.6%$10.0058/s5.63s
13Qwen3.8 Maxalibaba58.192.7%43.0%52.9%20.0%51.3%81.3%61.8%74.3%3.471.858.458.3%82.3%$3.0045/s2.56s
14Qwen3.8 2.4T A95Balibaba57.793.5%42.4%51.6%20.0%49.1%82.0%61.0%75.3%4.371.957.160.8%$3.0045/s2.53s
15GPT-5.6 Solhighopenai57.392.8%46.0%56.9%25.7%36.7%87.3%56.1%75.3%20.477.250.68.8%69.2%81.8%$11.2573/s9.55s
16Muse Spark 1.2xhighmeta56.890.4%45.5%56.4%17.7%34.8%80.1%56.4%83.3%27.272.249.366.7%$2.00
17GPT-5.6 Terramaxopenai56.692.5%42.9%53.9%30.0%40.2%88.0%53.8%79.7%0.176.750.212.1%71.2%80.7%$4.50116/s172.10s
18Gemini 3.7 Flashhighgoogle56.094.5%47.9%56.8%14.3%32.8%85.8%51.6%80.0%26.576.145.135.5%85.5%$1.50394/s14.52s
19Grok 4.5highxai55.893.1%42.7%54.1%15.4%42.1%81.6%51.2%74.0%25.372.448.945.9%80.4%$3.0064/s6.69s
20GPT-5.6 Solmediumopenai55.692.6%42.2%56.5%22.9%36.5%86.1%52.5%74.3%19.476.347.99.2%69.6%81.4%$11.2573/s3.40s
21Claude Sonnet 5adaptive · maxanthropic55.391.1%41.3%53.6%16.9%37.3%80.5%54.8%77.0%16.471.549.760.6%77.3%$4.0081/s129.13s
22Gemini 3.7 Flashmediumgoogle53.492.1%39.0%57.9%9.4%35.5%78.3%49.9%81.0%23.771.545.134.1%84.7%$1.50333/s5.84s
23DeepSeek V4 Pro 0813on · maxdeepseek53.292.8%41.0%49.2%18.0%39.6%78.7%54.5%75.3%0.868.849.65.2%$1.9881/s1.67s
24GPT-5.6 Terraxhighopenai52.890.8%41.9%51.6%27.1%29.7%80.1%53.6%75.0%-3.070.646.511.0%66.3%79.5%$4.50107/s21.50s
25GLM-5.2maxzai52.689.5%41.1%50.5%20.9%34.6%77.9%50.3%76.7%4.468.845.773.7%73.3%$2.1595/s2.04s
26Claude Opus 5adaptive · lowanthropic52.588.9%43.4%48.0%23.1%30.3%76.4%47.9%77.0%28.666.942.137.8%79.8%$10.0059/s4.15s
27GPT-5.6 Lunamaxopenai52.391.1%39.5%52.5%20.6%31.1%80.9%53.9%78.3%-10.371.446.97.4%78.6%$0.45157/s101.26s
28Qwen3.8 27Bxhighalibaba52.090.5%33.9%44.7%5.4%48.0%79.8%52.3%77.3%-10.068.150.969.7%76.3%$1.1357/s2.98s
29DeepSeek V4 Flash 0731on · maxdeepseek51.890.8%38.6%49.9%16.6%39.4%78.7%52.9%74.3%-14.369.148.48.3%$0.66137/s1.15s
30Grok 4.6lowxai51.787.9%27.6%48.4%5.7%38.1%75.3%52.7%78.7%25.966.347.869.4%$3.0070/s11.36s
31Gemini 3.6 Flashhighgoogle51.692.8%40.8%52.7%10.6%29.9%77.5%46.1%79.0%22.169.240.544.4%83.2%$1.50228/s18.76s
32Gemini 3.7 Flashlowgoogle50.990.1%35.1%53.6%5.7%29.5%79.8%47.6%78.3%22.171.041.632.3%84.9%$1.50323/s1.04s
33GPT-5.6 Sollowopenai50.789.8%39.4%55.4%14.9%29.1%76.8%47.1%73.0%18.969.741.610.6%66.5%81.0%$11.2569/s2.83s
34GPT-5.6 Terrahighopenai50.189.6%38.5%50.1%22.9%28.7%75.7%50.6%73.3%-3.567.143.510.2%64.4%79.1%$4.50107/s2.36s
35GPT-5.6 Lunaxhighopenai50.189.5%37.0%50.0%20.6%28.7%77.9%51.3%73.3%-10.868.644.47.5%78.6%$0.45157/s39.86s
36Kimi K3lowkimi48.384.2%25.0%51.2%3.1%41.6%82.4%38.5%77.0%3.972.039.622.9%78.5%$6.0037/s2.93s
37Gemini 3.1 Pro Previewgoogle47.794.1%47.0%58.9%17.7%21.4%73.8%23.3%79.0%31.968.823.049.1%77.1%82.4%$4.50131/s44.70s
38Motif 3motif-technologies47.483.4%37.0%40.6%6.6%35.3%74.9%38.7%72.3%10.263.537.671.6%
39GPT-5.6 Lunahighopenai47.089.2%33.4%50.7%16.6%25.2%69.7%48.3%74.0%-12.063.341.07.6%77.6%$0.45136/s6.64s
40GPT-5.6 Terramediumopenai46.887.2%33.3%49.7%17.4%25.6%72.3%45.2%70.3%-5.164.739.110.2%62.2%76.8%$4.50107/s1.80s
41Gemini 3.5 Flashmediumgoogle46.7估算92.1%41.3%53.0%10.9%79.7%20.838.2%74.6%83.9%$3.38218/s21.21s
42GPT-5.3 Codexxhighopenai45.5估算91.5%42.5%53.2%16.9%78.3%10.910.8%75.4%78.5%$4.81134/s75.87s
43MiniMax-M3minimax45.492.9%39.0%45.4%3.7%15.3%65.2%44.3%80.3%1.458.636.181.6%82.9%78.6%$0.52106/s1.32s
44Motif 3 (Beta)motif-technologies45.3估算86.9%40.4%44.3%6.0%70.8%37.8%70.7%-17.862.050.9%
45DeepSeek V4 Proon · maxdeepseek45.388.8%37.5%50.0%12.9%30.1%64.0%40.3%70.0%-10.759.437.85.9%76.5%$0.5482/s1.78s
46Qwen3.8 27Bmediumalibaba44.584.5%14.1%38.1%0.0%47.4%65.2%51.0%76.3%-36.156.149.833.3%74.2%$1.1365/s3.03s
47DeepSeek V4 Proon · highdeepseek43.790.5%35.2%46.4%10.0%26.2%64.8%39.6%67.0%-10.658.735.311.3%71.3%$0.5476/s1.76s
48Kimi K2.7 Codekimi43.089.6%35.0%47.5%10.0%20.2%67.4%34.5%75.0%-10.260.830.317.6%63.1%$1.7149/s3.05s
49MiMo-V2.5-Proxiaomi42.986.6%35.7%50.2%4.0%9.9%65.2%38.3%77.7%3.360.229.575.3%79.9%$0.5458/s5.09s
50Qwen3.8 27Blowalibaba42.984.5%14.0%39.8%0.0%32.2%67.4%49.5%74.7%-26.758.243.747.1%73.8%$1.1366/s2.99s
51Claude Sonnet 5off · highanthropic42.680.0%19.0%48.6%1.1%15.7%75.3%43.6%66.0%-0.766.434.248.0%71.9%$4.0067/s2.08s
52Inklingxhighthinking-machines42.387.2%31.9%46.1%5.4%29.1%55.1%36.9%73.3%2.052.134.132.3%73.5%$1.7684/s1.78s
53Hy3tencent42.289.7%33.5%47.6%4.9%22.9%64.4%35.7%74.7%-18.558.831.425.9%$0.2473/s2.93s
54GPT-5.6 Soloffopenai41.979.0%16.7%47.1%5.1%19.6%74.2%43.9%57.0%1.165.136.07.2%71.9%$11.2567/s1.36s
55Nex-N2-Pronex41.789.2%33.7%41.8%8.6%17.5%67.8%37.4%76.3%-27.159.131.25.1%66.2%$1.00140/s1.58s
56Solar Pro 4upstage41.689.1%29.2%43.6%5.4%23.3%57.3%38.8%70.7%-0.852.733.675.6%$0.5247/s2.54s
57GPT-5.6 Terralowopenai41.384.3%29.2%49.2%9.4%18.8%62.5%37.7%68.7%-6.858.131.510.1%59.7%76.1%$4.50103/s1.38s
58Inkling Smallthinking-machines41.289.5%33.3%48.7%8.3%18.8%55.1%38.4%69.3%-8.952.931.937.0%74.0%$0.52123/s1.63s
59JT-4.1 Flash 236B A21Bchina-mobile39.9估算84.5%17.8%38.2%0.0%59.6%38.2%70.3%-8.852.456.3%64.1%
60Agnes 2.5 Pro Alphaagnes-ai39.787.6%33.6%42.2%10.9%12.4%67.0%33.8%73.0%-25.158.825.411.9%$0.56130/s2.12s
61Qwen3.7 Plusalibaba39.490.0%35.6%45.5%9.1%17.5%61.0%22.2%69.0%1.155.920.772.3%78.0%80.5%$0.7056/s2.18s
62GPT-5.6 Lunamediumopenai38.985.9%25.8%45.8%4.9%17.7%53.2%38.7%72.0%-13.250.731.89.1%75.8%$0.45151/s2.56s
63Nemotron 3 Ultra 550B A55Bonnvidia38.386.7%28.4%39.9%3.1%14.2%53.9%33.1%71.0%-0.449.327.570.3%81.4%$1.14146/s1.74s
64MiMo-V2.5xiaomi38.084.9%27.2%43.1%3.7%8.7%63.7%32.5%68.3%-9.856.824.468.1%67.1%75.4%$0.1860/s3.47s
65Ling 3.0 Flashinclusionai37.885.5%23.7%41.1%1.7%27.2%55.4%30.4%67.0%-17.950.629.355.9%$0.11407/s2.70s
66Qwen3.6 27Bonalibaba37.784.2%23.1%39.8%1.1%16.7%60.7%32.0%73.3%-20.053.727.550.7%67.6%74.6%$1.3560/s3.68s
67Gemini 3.5 Flash-Litegoogle37.483.8%18.8%40.9%0.0%17.5%53.6%32.0%74.7%5.249.327.265.6%79.0%$0.85407/s8.96s
68Solar Open2 250Bupstage37.485.7%28.5%45.6%5.7%21.6%44.2%31.1%68.3%-1.844.727.874.6%
69MiMo-V2-Omni-0327xiaomi37.3估算85.5%22.6%39.5%0.9%69.7%-15.159.9%67.3%73.9%
70Grok 4.3mediumxai36.9估算89.0%30.0%44.6%4.9%68.3%16.783.1%83.3%75.8%$1.56126/s15.24s
71Grok 4.3lowxai36.3估算84.3%18.4%41.9%0.6%67.7%13.982.6%81.0%72.8%$1.56120/s6.37s
72MiMo-V2-Omnixiaomi35.9估算82.8%22.1%36.7%1.1%71.7%-20.151.1%53.5%69.9%
73Gemini 3.5 Flashminimalgoogle35.8估算82.8%24.1%48.8%1.4%58.3%0.725.7%47.3%80.1%$3.38191/s0.83s
74Claude Sonnet 4.6off · lowanthropic35.1估算79.7%11.2%44.1%0.9%64.7%-2.139.2%42.4%69.2%$6.0044/s1.65s
75Muse Glimmerhighmeta35.183.5%22.0%43.6%2.6%23.5%51.7%22.7%80.0%-32.949.022.918.1%74.3%$0.58104/s0.79s
76A.X-K2sk-telecom35.085.7%29.6%38.5%8.9%16.1%39.0%30.5%66.0%-8.238.825.767.0%
77GLM-5.2offzai34.868.6%9.8%36.1%3.1%16.7%51.7%44.7%45.0%-6.646.535.166.3%$2.15108/s1.98s
78GPT-5.6 Terraoffopenai34.674.6%11.4%44.6%2.0%15.7%56.2%37.2%55.0%-23.252.330.15.0%66.7%$4.50104/s0.83s
79Qwen3.5 397B A17Bonalibaba34.389.3%29.0%42.0%1.7%13.4%51.3%23.2%72.7%-30.848.219.811.1%78.8%77.3%$1.3585/s2.23s
80MiMo-V2-Flash (Feb 2026)xiaomi34.0估算83.5%22.1%38.3%2.9%68.7%-19.151.6%71.8%
81G9v3-39A5Bai9star34.080.5%17.5%34.0%0.3%22.1%32.6%38.3%62.0%3.833.132.987.0%$0.00
82LongCat 2.0longcat34.078.0%33.7%35.4%2.6%13.2%50.2%26.5%62.7%-23.445.322.024.6%$1.3044/s2.65s
83GPT-5.6 Lunalowopenai33.983.5%19.8%45.6%2.6%12.8%43.4%32.8%65.3%-14.744.225.710.1%74.0%$0.45138/s2.25s
84KAT Coder Pro V2kwaikat33.785.5%16.1%38.3%0.0%4.5%70.0%20.3%70.0%-22.559.515.144.0%66.7%$0.52107/s1.46s
85Qwen3.5 122B A10Bonalibaba32.885.7%25.2%42.0%0.6%15.3%47.6%24.4%70.3%-41.545.721.312.9%75.7%75.0%$1.10138/s2.32s
86Qwen3.5 397B A17Boffalibaba32.7估算86.1%19.8%41.1%0.9%62.0%-37.917.3%51.6%52.7%$1.3586/s2.19s
87Qwen3.6 35B A3Bonalibaba32.184.1%22.2%35.8%0.3%9.3%44.9%27.8%66.7%-22.241.921.649.5%64.4%75.0%$0.84126/s2.09s
88DeepSeek V4 Prooffdeepseek31.9估算71.7%8.2%42.4%0.9%49.7%-29.912.2%45.8%$0.5479/s1.79s
89Ring-2.6-1Tinclusionai31.785.7%21.6%42.4%3.7%17.9%43.1%21.1%67.3%-37.742.820.114.7%44.6%$0.85137/s3.97s
90Qwen3.5 Omni Plusalibaba31.3估算82.6%14.9%40.5%0.6%58.0%-13.163.0%51.2%70.5%$1.5050/s2.51s
91Qwen3.6 27Boffalibaba31.382.9%15.1%37.3%0.9%9.3%51.3%30.6%64.3%-53.246.623.616.0%45.7%71.7%$1.3559/s3.75s
92o3openai31.1估算82.7%20.1%41.0%1.1%73.3%-15.611.9%71.4%70.1%$3.50107/s6.95s
93K-EXAONE 2.0 0803lg31.082.9%18.6%41.0%0.9%11.5%40.4%24.0%57.7%-6.640.619.777.4%
94Step 3.7 Flashstepfun30.980.9%21.4%40.0%2.3%12.0%39.3%25.9%69.7%-37.339.621.715.0%67.3%75.3%$0.44133/s2.96s
95Mistral Medium 3.5mistral30.474.8%13.8%39.6%0.0%15.1%50.6%21.7%65.3%-36.846.919.218.4%68.8%64.9%$3.00146/s2.32s
96Claude 4.5 Haikuonanthropic29.967.2%10.4%43.3%0.0%9.3%44.2%20.7%73.7%-4.443.916.572.7%54.3%58.6%$2.0094/s14.07s
97Gemma 4 31Bongoogle29.785.7%23.6%43.4%1.4%14.8%43.4%15.5%68.3%-47.943.414.415.0%75.6%73.4%$0.0036/s1.11s
98DeepSeek V4 Flashoffdeepseek29.3估算71.6%7.8%37.3%0.3%37.3%-44.05.5%47.2%$0.12
99GPT-5.5 Instant (June 2026)openai29.282.3%19.9%48.6%0.0%11.8%34.8%10.8%67.0%3.739.411.126.7%$11.25146/s1.13s
100JT-35B-Flashchina-mobile29.0估算82.9%6.4%29.1%0.0%63.7%-24.236.2%42.0%
101MiMo-V2.5-Prooffxiaomi28.4估算76.2%14.8%39.1%1.1%39.3%-37.810.8%42.7%$0.5451/s3.85s
102Qwen3.5 122B A10Boffalibaba28.282.7%15.9%35.6%0.9%10.3%47.2%19.5%61.7%-55.043.316.28.4%50.8%70.3%$1.10161/s2.37s
103GPT-5.6 Lunaoffopenai26.864.5%7.2%39.9%0.3%9.7%39.0%28.7%38.7%-24.939.322.225.0%60.3%$0.45138/s0.80s
104Doubao Seed Codebytedance-seed26.5估算76.4%14.1%40.7%0.3%69.3%-35.220.0%51.4%68.1%
105Gemma 4 26B A4Bongoogle26.179.2%19.3%40.0%0.0%12.0%39.0%13.4%61.7%-50.839.311.013.6%72.4%69.2%$0.18
106Nemotron 3 Super 120B A12Bonnvidia25.780.0%20.8%36.0%3.1%10.3%38.6%9.9%60.3%-41.537.78.813.0%71.5%$0.35143/s1.83s
107MiMo-V2-Flashoffxiaomi25.1估算65.6%8.6%25.9%0.0%61.8%17.0%35.0%-48.549.824.0%39.9%
108Grok 4.3offxai25.065.8%6.8%37.4%0.0%8.0%34.1%29.9%28.0%-33.135.223.025.5%47.6%64.8%$1.56121/s0.80s
109Qwen3.6 35B A3Boffalibaba24.681.7%13.9%1.3%0.0%5.4%41.6%25.9%60.0%-60.028.119.27.9%36.2%71.0%$0.84146/s2.08s
110Ling 3.0 Tinyinclusionai24.573.4%9.3%24.2%0.0%20.8%27.7%13.6%58.7%-19.426.516.069.5%$0.00173/s2.60s
111Qwen3.5 35B A3Boffalibaba24.381.9%13.4%29.3%0.6%4.9%40.8%14.8%60.3%-63.037.011.86.6%44.5%69.2%$0.69169/s2.07s
112Claude 4.5 Haikuoffanthropic24.1估算64.6%4.2%34.4%0.0%48.3%-7.674.3%42.0%55.1%$2.0087/s0.85s
113gpt-oss-120bhighopenai24.178.2%19.6%38.9%1.1%12.8%26.2%15.0%51.0%-49.330.413.49.2%69.0%$0.26163/s0.84s
114Nemotron 3.5 Lightningnvidia23.674.3%10.6%31.6%0.0%8.9%24.3%16.2%55.3%-17.726.813.862.4%$0.11314/s0.96s
115Command A+cohere22.876.1%12.0%37.8%0.3%6.0%22.8%10.8%48.7%-4.027.89.285.8%73.9%63.2%$0.00196/s0.39s
116K-EXAONEonlg22.5估算78.3%13.9%35.6%1.1%30.3%4.6%59.3%-58.032.111.2%64.7%
117ERNIE 5.0 Thinking Previewbaidu22.3估算77.7%13.3%37.5%1.4%7.0%-45.613.0%41.4%64.6%
118Gemma 4 31Boffgoogle22.376.3%11.8%41.1%0.0%8.9%29.2%12.4%42.7%-51.733.211.118.1%53.5%70.3%$0.2177/s1.37s
119Gemma 4 12Bongoogle22.2估算75.3%15.7%38.2%0.0%27.3%7.2%61.7%-52.731.019.0%73.5%69.7%$0.15112/s1.85s
120Nova 2.0 Pro Previewmediumaws22.1估算78.5%9.4%42.7%0.0%29.6%9.1%58.0%-49.034.09.0%79.0%64.5%$3.44118/s14.54s
121Mercury 2inception21.977.0%17.1%38.7%0.8%9.5%27.3%9.9%40.7%-50.731.19.58.8%69.8%$0.38924/s3.54s
122Qwen3.5 9Bonalibaba21.880.6%14.9%27.5%0.3%7.0%29.2%7.2%65.3%-53.528.77.016.4%66.7%69.2%$0.1586/s2.08s
123Qwen3 Coder Nextalibaba21.373.7%10.1%32.3%0.0%5.4%38.2%10.8%42.3%-62.436.28.96.3%35.2%$0.56122/s1.09s
124Nova 2.0 Omnimediumaws21.3估算76.0%7.0%36.2%0.0%56.3%-59.16.7%66.2%61.9%$0.85
125Apriel-v1.6-15B-Thinkerservicenow20.8估算73.3%10.8%37.3%0.3%-59.48.1%69.1%$0.00
126Nova 2.0 Litehighaws20.8估算81.1%11.6%36.9%0.3%16.1%4.5%58.7%-55.023.08.6%70.7%63.8%$0.85156/s12.62s
127Qwen3.5 9Boffalibaba20.6估算78.6%9.4%27.7%0.6%21.3%45.0%-70.723.51.6%37.8%66.8%$0.1985/s0.77s
128EXAONE 4.5 33Blg20.5估算79.4%12.9%28.0%0.3%21.3%9.0%51.7%-50.723.620.1%58.0%67.3%
129Gemma 4 26B A4Boffgoogle20.4估算71.4%11.5%37.3%0.0%40.7%-62.37.9%45.4%66.7%$0.2076/s1.27s
130Qwen3.5 4Bonalibaba20.4估算77.1%9.9%16.1%0.0%6.8%25.8%61.0%-58.422.613.4%52.0%65.4%$0.0652/s0.82s
131North Mini Codecohere20.275.7%11.1%38.2%0.3%6.4%35.6%2.1%36.0%-48.636.53.116.8%57.6%$0.0027/s0.53s
132Nova 2.0 Pro Previewlowaws19.8估算75.1%5.2%38.7%0.0%19.5%7.6%63.3%-40.525.930.5%79.6%62.7%$3.44115/s9.48s
133Mistral Small 4onmistral19.776.9%9.9%38.0%0.3%4.9%21.0%4.5%47.3%-30.426.64.633.5%48.2%56.8%$0.26149/s0.80s
134Devstral 2mistral19.259.4%3.6%33.1%0.0%10.5%30.3%12.2%31.3%-46.731.310.614.7%38.1%$0.0042/s1.96s
135Nova 2.0 Litemediumaws19.2估算76.8%9.0%36.8%0.0%60.3%-57.08.2%68.5%62.5%$0.85152/s14.17s
136Qwen3.5 Omni Flashalibaba19.2估算74.2%7.6%25.5%0.0%48.7%-65.16.6%38.0%64.7%$0.28264/s1.84s
137JT-MINIchina-mobile18.8估算67.6%6.4%27.2%0.0%12.7%-64.18.7%36.7%
138Trinity Large Thinkingarcee18.775.2%15.8%36.1%0.9%5.8%20.6%3.1%38.3%-44.125.83.714.1%56.3%$0.40198/s1.13s
139HyperNova 60B 2605multiversecomputing18.373.3%16.6%33.0%0.9%5.6%18.4%7.9%35.3%-54.623.26.817.7%66.5%
140Magistral Medium 1.2mistral18.073.9%10.3%39.2%0.3%5.6%12.4%0.0%53.3%-26.721.31.940.0%43.0%59.7%
141Nova 2.0 Litelowaws18.0估算69.8%4.0%33.3%0.0%55.0%-54.112.9%61.2%58.0%$0.85158/s9.49s
142Nemotron Cascade 2 30B A3Bnvidia17.8估算75.8%12.0%34.8%0.6%20.6%0.1%38.3%-51.525.314.6%80.4%
143Devstral Small 2mistral17.753.2%3.5%28.8%0.0%10.7%29.6%11.6%27.0%-57.729.310.812.7%31.2%44.6%$0.00135/s2.31s
144K2 Think V2mbzuai17.4估算71.3%10.1%33.0%0.0%15.0%0.0%55.3%-40.321.028.8%62.8%
145LongCat Flash Litelongcat17.4估算63.6%5.8%28.4%0.0%28.7%-70.13.8%43.1%
146HyperCLOVA X SEED Think (32B)naver17.2估算61.5%5.5%28.4%0.0%13.3%-53.318.8%37.9%
147K-EXAONEofflg16.9估算69.5%5.7%27.0%0.0%50.0%-65.48.5%39.6%
148Qwen3 Next 80B A3Bonalibaba16.975.9%12.6%38.8%0.0%6.2%6.7%0.0%62.0%-51.417.42.113.0%60.7%$0.41208/s2.27s
149Nova 2.0 Omnilowaws16.7估算69.9%34.3%0.0%54.0%-50.315.3%61.8%59.8%$0.85
150korea-telecomKTMi:dm K 2.5 Prokorea-telecom16.6估算70.1%8.1%33.2%0.0%10.0%-59.26.1%49.3%
151G9v3-3Bai9star16.2估算43.8%4.5%17.7%0.0%6.0%18.3%40.7%-4.49.988.3%$0.00
152Qwen3.5 4Boffalibaba16.1估算71.2%8.0%18.3%0.0%4.3%21.3%34.0%-74.720.33.2%33.3%62.1%$0.0619/s0.90s
153Mistral Large 3mistral15.968.0%4.2%36.2%0.0%5.8%12.0%7.0%34.7%-39.620.15.514.0%36.2%55.7%$0.7543/s1.36s
154INTELLECT-3prime-intellect15.7估算76.1%13.1%39.1%0.3%36.3%-50.912.4%34.0%
155Solar Open 100Bonupstage15.2估算65.7%10.4%26.9%0.0%37.0%-54.510.4%57.7%
156gpt-oss-20bhighopenai15.268.8%11.0%34.4%1.4%7.0%13.9%3.3%33.3%-63.020.73.15.9%65.1%$0.09113/s1.14s
157Nemotron 3 Nano Omni 30B A3B Reasoningnvidia15.0估算46.9%4.8%27.8%0.0%6.7%0.0%40.7%-57.413.814.3%63.2%53.2%$0.13326/s1.00s
158gpt-oss-120blowopenai14.967.2%5.9%36.0%0.0%2.9%13.9%0.0%46.0%-53.521.21.08.6%58.3%$0.26187/s0.88s
159NVIDIA Nemotron 3 Nano 30B A3Bonnvidia14.575.7%11.4%29.6%0.9%6.0%6.7%0.0%37.3%-51.614.42.016.7%71.1%$0.09258/s1.12s
160Solar Pro 3upstage14.572.4%10.3%24.7%0.0%8.7%12.0%0.0%31.0%-53.416.22.911.8%71.2%$0.26146/s2.30s
161Llama 4 Maverickmeta14.567.1%4.9%33.1%0.0%3.7%7.9%0.0%50.0%-41.916.31.211.1%43.0%62.1%$0.4294/s0.92s
162Nova 2.0 Pro Previewoffaws14.4估算63.6%3.9%28.1%0.0%17.2%3.0%30.7%-49.520.920.1%52.0%$3.4498/s1.05s
163gpt-oss-20blowopenai14.4估算61.1%5.3%34.0%0.0%30.7%-58.513.3%57.8%$0.10116/s1.19s
164K2-V2highmbzuai14.2估算68.1%10.5%28.6%0.0%36.3%-55.69.0%60.1%
165Qwen3 Next 80B A3B Instructalibaba13.8估算73.8%7.6%30.7%0.0%52.0%-59.57.3%39.7%$0.41197/s2.13s
166DiffusionGemma 26B A4Bgoogle13.5估算66.9%10.8%34.3%0.3%12.4%2.7%18.3%-59.219.79.1%59.5%66.5%
167Gemma 4 12Boffgoogle13.2估算66.1%6.3%29.7%0.0%31.3%-52.826.6%45.2%62.0%$0.15125/s2.31s
168Motif-2-12.7B-Reasoningmotif-technologies12.8估算69.5%8.8%28.2%0.0%13.3%-61.79.4%57.0%
169Nova Premieraws12.7估算56.9%4.2%27.9%0.0%31.3%-35.428.1%36.2%$5.0032/s2.85s
170K2-V2mediummbzuai12.4估算59.8%4.4%25.2%0.0%28.3%-49.018.4%55.1%
171Llama Nemotron Super 49B v1.5onnvidia12.4估算74.8%7.3%34.8%0.0%35.0%-47.621.8%37.0%$0.40125/s4.34s
172Celeris-1celeris12.463.1%6.8%20.7%0.0%3.9%11.2%1.7%37.3%-71.614.42.47.2%$0.331573/s0.58s
173Mistral Small 4offmistral12.3估算57.1%3.8%28.1%0.3%24.3%-48.522.0%32.8%46.5%$0.26140/s0.82s
174Tri-21B-Thinktrillionlabs12.3估算60.1%5.9%17.4%0.3%15.0%-62.514.6%54.6%
175Gemma 4 E4Bongoogle12.2估算57.6%3.8%24.4%0.6%1.9%0.0%33.0%-19.79.469.1%44.2%51.4%$0.0456/s0.70s
176MiniCPM5-1Bonopenbmb11.9估算27.8%6.5%4.4%0.0%4.7%-14.683.0%49.3%
177Sarvam 105Bhighsarvam11.9估算73.8%11.0%26.4%0.0%0.0%-59.46.6%34.4%$0.07
178Nova 2.0 Liteoffaws11.8估算60.3%2.9%24.0%0.0%18.3%-59.814.1%40.5%49.0%$0.85149/s1.07s
179MiniCPM5-1Boffopenbmb11.7估算26.9%4.5%1.4%0.0%4.7%-0.899.1%35.2%
180Magistral Small 1.2mistral11.566.3%6.4%35.2%0.3%4.5%4.5%0.0%18.0%-65.114.71.59.2%44.4%55.5%$0.75
181Ministral 3 14Bmistral11.257.2%4.6%23.6%0.0%6.6%9.7%0.0%25.0%-66.414.42.27.5%32.0%49.8%$0.2084/s1.00s
182Nanbeige4.1-3Bnanbeige11.0估算84.9%10.9%26.6%0.0%1.1%0.0%0.0%-42.59.641.9%35.4%
183Unknown developerUDLFM2.5-2.6B未匹配11.055.8%6.2%14.2%0.0%7.2%4.5%0.0%5.3%-10.97.72.484.0%$0.00
184EXAONE 4.0 32Bonlg10.5估算73.9%11.4%34.4%0.0%14.7%-60.114.2%36.3%
185Nova 2.0 Omnioffaws10.4估算55.5%3.8%27.9%0.0%23.7%-64.210.7%41.1%49.9%$0.85
186Llama 4 Scoutmeta10.358.7%3.8%17.0%0.0%3.3%3.7%0.0%30.3%-52.18.21.120.6%39.5%52.9%$0.30133/s0.79s
187Hermes 4 - Llama-3.1 70Bonnous-research9.9估算69.9%8.8%34.1%0.0%8.0%-49.44.6%31.3%$0.2077/s1.47s
188Falcon-H1R-7Btii-uae9.7估算66.1%11.0%24.9%0.3%11.3%-61.311.2%54.4%
189Gemma 4 E2Bongoogle9.5估算43.3%4.8%20.9%0.0%0.4%0.0%17.0%-23.67.267.6%38.0%44.6%
190Qwen3 Omni 30B A3Bonalibaba9.5估算72.6%7.5%30.6%0.0%0.0%-61.411.0%43.4%60.2%$0.43100/s1.95s
191Step3 VL 10Bstepfun9.3估算69.0%10.8%31.1%0.0%0.0%-59.016.7%50.2%64.0%
192Llama 3.3 70B Instructmeta9.3估算49.8%3.6%26.0%0.0%4.9%0.0%15.7%-54.211.99.8%47.1%$0.6781/s1.63s
193Ministral 3 8Bmistral9.047.1%4.3%20.8%0.0%3.7%4.1%0.0%25.3%-69.09.71.25.8%29.1%46.0%$0.15114/s0.74s
194Llama 3.1 Nemotron Ultra 253B v1onnvidia8.9估算72.8%7.4%34.7%0.0%7.7%-44.918.8%38.2%$0.9052/s2.34s
195ERNIE 4.5 300B A47Bbaidu8.9估算81.1%3.3%31.5%0.0%4.0%-35.133.1%39.1%$0.49
196Hermes 4 - Llama-3.1 405Bonnous-research8.8估算72.7%10.9%25.2%0.3%22.7%-36.05.5%32.7%$1.5033/s2.57s
197NVIDIA Nemotron Nano 12B v2 VLonnvidia8.8估算57.2%5.5%26.2%0.0%42.0%-64.27.7%31.9%52.9%$0.30102/s3.59s
198Gemma 4 E4Boffgoogle8.7估算54.9%4.8%3.9%0.3%22.0%-41.046.2%40.5%51.2%$0.0458/s0.65s
199Granite 4.1 30Bibm8.7估算48.1%4.1%25.8%0.0%2.6%0.0%20.3%-67.810.45.0%44.4%
200NVIDIA Nemotron Nano 9B V2onnvidia8.7估算57.0%4.9%22.0%0.0%23.0%-41.539.5%27.6%$0.07183/s4.07s
201Hermes 4 - Llama-3.1 405Boffnous-research8.6估算53.6%4.2%34.6%0.0%21.0%-33.319.8%34.8%$1.5034/s2.37s
202NVIDIA Nemotron 3 Nano 4Bnvidia8.6估算51.3%4.9%16.4%0.0%3.7%0.0%17.3%-70.58.011.4%58.2%
203Llama Nemotron Super 49B v1.5offnvidia8.5估算48.1%4.3%23.8%0.0%24.3%-46.232.7%32.9%$0.40122/s4.99s
204K2-V2lowmbzuai8.4估算54.1%3.6%22.3%0.0%20.0%-47.324.0%41.0%
205Kimi Linear 48B A3B Instructkimi8.4估算41.2%2.5%19.9%26.7%28.1%
206Llama 3.1 Instruct 405Bmeta8.3估算51.5%4.0%29.9%0.0%24.7%-17.147.6%39.0%
207LFM2.5-8B-A1Bliquidai8.1估算51.3%6.9%7.8%0.0%0.0%-33.253.1%55.6%$0.00339/s1.46s
208Ring-flash-2.0inclusionai8.0估算72.5%9.6%16.8%0.3%23.7%-58.69.9%43.3%$0.25
209Olmo 3.1 32B Thinkai27.9估算59.1%6.3%29.3%0.0%0.0%-43.433.7%66.0%$0.00
210Command Acohere7.5估算52.7%4.0%28.1%0.0%20.0%-48.322.7%36.5%$4.3869/s1.58s
211Qwen3.5 2Bonalibaba7.4估算45.6%2.6%2.8%0.0%3.0%0.0%28.7%-60.22.925.8%31.5%43.0%
212Llama 3.1 Nemotron Instruct 70Bnvidia7.4估算46.5%4.2%23.3%0.0%7.3%-40.828.7%30.7%$1.20110/s3.35s
213NVIDIA Nemotron 3 Nano 30B A3Boffnvidia7.2估算39.9%4.6%23.0%0.0%0.0%8.7%-69.39.0%37.5%$0.09230/s0.92s
214NVIDIA Nemotron Nano 9B V2offnvidia7.2估算55.7%4.6%20.9%0.0%26.0%-57.325.6%27.1%$0.09169/s1.27s
215Ministral 3 3Bmistral7.135.8%5.4%14.4%0.0%4.7%0.0%0.0%16.0%-64.04.81.619.8%26.8%38.1%$0.10204/s0.63s
216Hermes 4 - Llama-3.1 70Boffnous-research6.7估算49.1%3.6%27.7%0.0%3.0%-48.617.3%29.0%$0.2080/s1.36s
217Granite 4.1 8Bibm6.4估算43.3%3.8%21.8%0.0%3.4%11.3%-64.19.512.9%38.6%$0.06112/s0.77s
218Sarvam 30Bhighsarvam6.4估算63.3%7.5%19.2%0.3%0.0%-71.53.7%26.5%$0.05
219Olmo 3.1 32B Instructai26.2估算53.9%5.0%16.7%0.0%0.0%-52.426.8%39.2%
220Gemma 4 E2Boffgoogle6.2估算40.5%4.7%20.4%0.0%14.7%-60.225.6%33.6%41.8%
221R1 1776perplexity6.0估算
222Llama 3.2 Instruct 90B (Vision)meta6.0估算43.2%4.5%24.0%39.5%
223Phi-4 Mini Instructazure5.7估算33.1%4.5%10.8%0.0%0.4%0.0%16.7%-61.13.822.0%21.1%$0.0043/s0.85s
224EXAONE 4.0 32Bofflg5.7估算62.8%5.0%25.2%0.0%9.0%-62.817.9%33.5%
225Qwen3.5 2Boffalibaba5.3估算43.8%5.0%7.2%0.0%0.0%0.0%15.0%-83.12.42.3%29.1%42.7%
226Qwen3.5 0.8Bonalibaba5.2估算11.1%1.1%0.0%0.0%0.0%0.0%11.0%-54.50.038.7%21.5%25.8%
227DeepHermes 3 - Mistral 24B Previewoffnous-research5.0估算38.2%3.8%22.8%
228Jamba 1.7 Largeai21-labs5.0估算39.0%3.7%18.8%0.0%18.3%-54.94.9%35.2%
229Granite 4.0 H Smallibm4.9估算41.6%3.8%20.9%0.0%11.0%-61.012.2%31.5%$0.1125/s26.74s
230Qwen3 Omni 30B A3B Instructalibaba4.8估算62.0%4.6%18.6%0.0%0.0%-69.32.4%31.2%55.5%$0.4389/s1.86s
231LFM2 24B A2Bliquidai4.6估算47.4%4.2%10.9%0.0%0.0%-58.131.0%45.9%
232microsoftMiPhi-4microsoft4.6估算57.5%3.8%26.0%0.0%0.0%-55.718.8%23.5%$0.2240/s2.45s
233Nova Microaws4.4估算35.8%4.6%9.4%0.0%10.7%-48.534.9%29.4%$0.06273/s0.90s
234Granite 4.1 3Bibm4.4估算31.4%3.4%11.9%0.0%1.1%0.0%3.3%-77.04.74.9%33.7%
235NVIDIA Nemotron Nano 12B v2 VLoffnvidia4.2估算43.9%4.3%17.6%0.0%19.7%-72.24.5%25.9%44.5%$0.30148/s1.71s
236Phi-4 Multimodal Instructazure4.2估算31.5%5.0%11.0%14.5%$0.0018/s0.86s
237MiniCPM-V 4.6 1.3Bopenbmb3.8估算30.5%5.1%2.1%0.0%0.0%0.0%6.7%-82.90.73.9%26.7%37.9%
238Jamba Reasoning 3Bai21-labs3.8估算33.3%3.8%5.9%0.0%8.3%-57.731.0%52.4%
239Reka Flash 3reka-ai3.7估算52.9%4.4%26.7%0.0%0.0%-65.18.7%30.4%$0.35
240Olmo 3 7B Thinkai23.6估算51.6%6.0%21.2%0.0%0.0%-71.86.8%41.5%
241Molmo 7B-Dai23.4估算24.0%5.1%3.6%0.0%19.7%24.5%
242Llama 3.2 Instruct 11B (Vision)meta3.0估算22.1%5.5%11.2%0.0%15.7%-62.618.1%30.4%29.3%$0.3425/s1.17s
243Qwen3.5 0.8Boffalibaba2.9估算23.6%5.1%2.9%0.0%0.4%0.0%9.0%-88.61.21.5%21.6%25.7%
244Exaone 4.0 1.2Bonlg2.5估算51.5%6.0%9.3%0.0%0.0%-79.87.0%23.0%
245Olmo 3 7B Instructai22.4估算40.0%5.8%10.3%0.0%0.0%-77.18.8%32.8%$0.13
246Exaone 4.0 1.2Bofflg2.4估算42.4%5.7%7.4%0.0%0.0%-82.28.3%25.3%
247LFM2.5-1.2B-Thinkingliquidai2.3估算33.9%6.2%4.2%0.0%0.0%-80.05.6%41.8%
248Jamba 1.7 Miniai21-labs2.3估算32.2%4.4%9.3%0.0%15.3%-73.14.9%31.4%
249LFM2 2.6Bliquidai2.3估算30.6%5.5%2.5%0.0%0.0%-54.137.1%19.5%
250LFM2.5-1.2B-Instructliquidai2.3估算32.6%6.7%2.3%0.0%0.0%-72.114.9%43.8%
251Granite 4.0 H 1Bibm2.2估算26.3%5.0%8.2%0.0%6.0%-72.318.3%26.2%
252Gemma 3 270Mgoogle2.0估算22.4%3.6%0.0%0.0%0.0%-29.369.6%12.1%
253Apertus 70B Instructswiss-ai-initiative2.0估算27.2%5.5%5.7%0.0%0.0%-61.517.2%25.9%$1.34
254Granite 4.0 Microibm2.0估算33.6%5.0%11.9%0.0%5.7%-77.54.2%24.8%
255DeepHermes 3 - Llama-3.1 8B Previewoffnous-research1.9估算27.0%4.3%9.1%
256Granite 4.0 1Bibm1.6估算28.1%4.8%8.7%0.0%6.0%-81.66.5%20.5%
257Molmo2-8Bai21.6估算42.5%4.3%13.3%0.0%0.0%-69.28.9%26.9%37.5%
258LFM2 8B A1Bliquidai1.3估算34.4%4.9%6.8%0.0%1.0%-76.57.7%26.3%
259Apertus 8B Instructswiss-ai-initiative1.0估算25.6%5.0%4.1%0.0%0.0%-77.23.9%22.4%$0.13
260Granite 4.0 350Mibm1.0估算26.1%5.5%0.9%0.0%0.0%-69.323.9%15.9%
261Granite 4.0 H 350Mibm1.0估算25.7%6.4%1.7%0.0%0.0%-80.911.9%17.6%
262LFM2.5-VL-1.6Bliquidai1.0估算28.9%5.1%3.0%0.0%0.0%-84.44.3%33.1%26.5%$0.00346/s1.13s
263Tiny Aya Globalcohere1.0估算30.5%5.2%3.6%0.0%0.0%-84.33.8%20.1%$0.00
Claude Sonnet 5adaptive · highanthropic45.1%$4.0067/s8.95s
Claude Sonnet 5adaptive · lowanthropic36.0%$4.0066/s2.41s
Claude Sonnet 5adaptive · mediumanthropic40.2%$4.0066/s3.79s
Claude Sonnet 5adaptive · xhighanthropic50.1%$4.0073/s25.76s
Cogito v2.1ondeepcogito76.8%12.0%41.0%0.0%22.0%-25.819.8%46.3%$1.25
EXAONE 4.5 33Bofflg
GPT-5.5 Proxhighopenai30.6%
Gemini 3 Deep Thinkgoogle25.7%

各列衡量的是什么

Artificial Analysis Intelligence Index
越高越好
GPQA Diamond
越高越好 · 推理与知识
Humanity's Last Exam
越高越好 · 推理与知识
SciCode
越高越好 · 编程
CritPt
越高越好 · 推理与知识
τ³-Banking
越高越好 · 智能体工作
Terminal-Bench 2.1
越高越好 · 智能体编程
GDPval-AA
越高越好 · 智能体工作
AA-LCR
越高越好 · 长上下文
AA-Omniscience Index
越高越好 · 事实性
Coding Index
越高越好 · 编程
Agentic Index
越高越好 · 智能体编程
AA-Omniscience Non-Hallucination Rate
越高越好 · 事实性
IFBench
越高越好 · 指令遵循
MMMU-Pro
越高越好 · 多模态
Blended Price, USD per 1M tokens (3:1 input:output)
越低越好 · 成本
Output Speed
越高越好 · 速度
Time To First Token
越低越好 · 速度