9 项评测

智能指数拆分

拆解模型在知识、推理、编程、工具使用和长上下文上的差异。

数据快照:2026-07-21查看原始来源 ↗
主要指标Evaluation relevance
快照日期2026-07-21

INTELLIGENCE INDEX V4.1

当前 9 项组成评测

01 · Agent

GDPval-AA v2

高经济价值、可交付成果导向的真实知识工作任务。

02 · Agent

τ³-Banking

多轮客户交互、知识库检索与工具调用。

03 · 编码

Terminal-Bench v2.1

在真实终端环境中完成构建、配置、调试和运维任务。

04 · 编码

SciCode

科研场景中的代码生成、执行和数值计算。

05 · 通用

AA-LCR

跨长上下文定位、保持和综合关键信息。

06 · 通用

AA-Omniscience

同时衡量知识准确性、拒答判断和幻觉风险。

07 · 科学推理

Humanity's Last Exam

覆盖多学科专家级知识与复杂推理。

08 · 科学推理

GPQA Diamond

研究生水平、难以通过简单检索回答的科学问题。

09 · 科学推理

CritPt

要求严谨分析与多步骤推理的科学评测。

完整数据

逐模型 9 项评测成绩

直接展示来源页面公开数据中的全部模型,不截断为精选榜单。

447 / 447 个模型

智能指数带 * 为原站估算值;其余百分比列已统一换算为百分数。GDPval-AA v2 与 AA-Omniscience 保留原始分值口径。

Artificial Analysis Intelligence Index v4.1 完整逐模型分项成绩
#模型机构智能指数GDPval-AA v2τ³-BankingTerminal-Bench v2.1SciCodeAA-LCRAA-OmniscienceHLEGPQA DiamondCritPt权重推理发布日期
1Claude Fable 5Anthropic59.861,759.626.8%84.6%60.2%70%40.1553.3%92.6%28.6%闭源2026-06-09
2GPT-5.6 Sol (max)OpenAI58.891,742.6133%88%56.1%73.7%21.747.2%94.1%32.3%闭源2026-07-09
3GPT-5.6 Sol (xhigh)OpenAI57.651,702.2132.6%89.5%56%71%20.5544.7%93.1%28.6%闭源2026-07-09
4Kimi K3Kimi57.111,679.2133.4%85%58.7%74.7%18.4244.3%93.5%23.4%闭源2026-07-16
5GPT-5.6 Sol (high)OpenAI55.871,630.2930.6%87.3%56.9%68.3%19.7544.1%92.8%25.7%闭源2026-07-09
6Claude Opus 4.8 (max)Anthropic55.691,600.1227.6%84.6%53.5%67.7%27.4345.7%92%20.9%闭源2026-05-28
7GPT-5.6 Terra (max)OpenAI54.951,581.2331.8%88%53.9%74%-0.2241.8%92.5%30%闭源2026-07-09
8GPT-5.5 (xhigh)历史OpenAI54.841,493.1831.3%84.3%56.1%74.3%20.0744.3%93.5%27.1%闭源2026-04-23
9Grok 4.5 (high)SpaceXAI53.831,53532.6%81.6%54.1%67.7%26.3840.3%93.1%15.4%闭源2026-07-08
10GPT-5.6 Sol (medium)OpenAI53.591,561.6526.5%86.1%56.5%68.7%18.9539.7%92.6%22.9%闭源2026-07-09
11Claude Opus 4.7 (max)历史Anthropic53.531,499.6428.9%83.1%54.5%70.3%26.1739.6%91.4%12%闭源2026-04-16
12Claude Sonnet 5 (max)Anthropic53.351,60728.2%80.5%53.6%70.7%15.3239.6%91.1%16.9%闭源2026-06-30
13GPT-5.5 (high)历史OpenAI53.131,471.0229.5%79.4%55.9%73.3%18.3743%93.2%25.4%闭源2026-04-23
14GPT-5.6 Terra (xhigh)OpenAI51.61,571.9724.3%80.1%51.6%71.3%-3.340%90.8%27.1%闭源2026-07-09
15GPT-5.4 (xhigh)历史OpenAI51.41,394.6830.3%78.3%56.6%74%5.6541.6%92%23.4%闭源2026-03-05
16GPT-5.6 Luna (max)OpenAI51.241,58427.2%80.9%52.5%74%-11.2337.2%91.1%20.6%闭源2026-07-09
17GLM-5.2 (max)Z AI51.091,513.8926.8%77.9%50.5%71.3%3.9740.1%89.5%20.9%开放2026-06-16
18Muse Spark 1.1 (xhigh)Meta50.621,374.3325.2%77.9%58.2%63.3%1845.1%89.8%15.1%闭源2026-07-09
19GPT-5.5 (medium)历史OpenAI50.411,374.6325.8%80.5%53.5%72.3%17.2840.6%92.6%18.6%闭源2026-04-23
20Gemini 3.5 FlashGoogle50.21,34925.4%78.7%53.1%69.3%22.6841%92.2%13.1%闭源2026-05-19
21GPT-5.6 Sol (low)OpenAI49.441,444.6824.4%76.8%55.4%67.7%18.436.6%89.8%14.9%闭源2026-07-09
22GPT-5.6 Luna (xhigh)OpenAI49.071,539.1124.3%77.9%50%69.7%-11.635.6%89.5%20.6%闭源2026-07-09
23GPT-5.6 Terra (high)OpenAI48.951,512.8422.3%75.7%50.1%72.3%-4.0736.7%89.6%22.9%闭源2026-07-09
24Claude Sonnet 4.6 (max)历史Anthropic47.211,377.1630.5%71.2%46.8%70.7%12.3730%87.5%3.1%闭源2026-02-17
25Gemini 3.1 Pro PreviewGoogle46.4696516.5%73.8%58.9%72.7%32.9344.7%94.1%17.7%闭源2026-02-19
26GPT-5.6 Luna (high)OpenAI46.061,472.3622.3%69.7%50.7%69%-12.4531.6%89.2%16.6%闭源2026-07-09
27Qwen3.7 MaxAlibaba45.991,273.1510.9%74.5%48.8%69%14.0838.1%92.3%13.4%闭源2026-05-19
28GPT-5.6 Terra (medium)OpenAI45.571,404.4519.4%72.3%49.7%68%-5.1731.6%87.2%17.4%闭源2026-07-09
29Gemini 3.5 Flash (medium)Google45.38*53%71%21.6539.9%92.1%10.9%闭源2026-05-19
30MiniMax-M3MiniMax44.441,394.8313%65.2%45.4%74%1.3737.1%92.9%3.7%开放2026-06-01
31GPT-5.3 Codex (xhigh)OpenAI44.27*53.2%74%9.8839.9%91.5%16.9%闭源2026-02-05
32DeepSeek V4 Pro (max)DeepSeek44.271,307.2125.8%64%50%66.3%-10.0235.9%88.8%12.9%开放2026-04-24
33Kimi K2.6历史Kimi44.221,189.420.6%65.9%53.5%69.7%6.4235.9%91.1%8%开放2026-04-20
34Motif 3 (Beta)Motif Technologies44.081,259.0428.9%70.8%44.3%61.3%-17.2838.2%86.9%6%闭源
35Claude Opus 4.6 (max)历史Anthropic43.71*51.9%70.7%13.536.7%89.6%12.6%闭源2026-02-05
36GPT-5.5 (low)历史OpenAI43.461,190.8421.2%65.5%51.6%72%14.6231%91%8%闭源2026-04-23
37DeepSeek V4 Pro (high)DeepSeek43.111,298.7623.7%64.8%46.4%65%-9.733.5%90.5%10%开放2026-04-24
38Muse SparkMeta43.061,143.6919.6%62.2%51.5%69.7%4.0839.9%88.4%11.3%闭源2026-04-08
39Claude Opus 4.7 (Non-reasoning, high)Anthropic42.68*50.1%67%14.2231.2%88.5%5.1%闭源2026-04-16
40MiMo-V2.5-ProXiaomi42.241,265.48.7%65.2%50.2%73.3%3.633.8%86.6%4%开放2026-04-22
41GPT-5.2 (xhigh)历史OpenAI42.18*52.1%72.7%-135.4%90.3%11.6%闭源2025-12-11
42Kimi K2.7 CodeKimi41.951,186.7918.1%67.4%47.5%66.3%-10.732.8%89.6%10%开放2026-06-12
43Claude Sonnet 5 (Non-reasoning)Anthropic41.731,371.8514%75.3%48.6%58.7%-1.1817.8%80%1.1%闭源2026-06-30
44Hy3Tencent41.231,213.8720.8%64.4%47.6%66.7%-18.531.6%89.7%4.9%开放2026-07-06
45GPT-5.6 Sol (Non-reasoning)OpenAI41.21,383.9316.1%74.2%47.1%55.3%0.8515.8%79%5.1%闭源2026-07-09
46Nex-N2-ProNex AGI40.961,256.8716.9%67.8%41.8%67.7%-27.6232.4%89.2%8.6%开放2026-06-02
47Claude Opus 4.5历史Anthropic40.77*49.5%74%13.2728.4%86.6%4.6%闭源2025-11-24
48InklingThinking Machines40.741,238.8523.7%55.1%46.1%63.3%2.0529.7%87.2%5.4%开放2026-07-15
49GPT-5.6 Terra (low)OpenAI40.471,248.5416.1%62.5%49.2%64.3%-7.1227.4%84.3%9.4%闭源2026-07-09
50MiMo-V2-Pro历史Xiaomi40.29*42.5%60.7%4.9228.3%87%0.3%闭源2026-03-18
51DeepSeek V4 Flash (max)DeepSeek40.281,188.5422.9%61.8%44.9%63%-22.932.1%89.4%7.1%开放2026-04-24
52GLM-5.1历史Z AI40.161,256.6811.5%61.8%43.8%62.3%1.9328%86.8%4.6%开放2026-04-07
53GPT-5.2 Codex (xhigh)历史OpenAI40.14*54.6%75.7%-2.4833.5%89.9%8.7%闭源2025-12-11
54Qwen3.6 Max Preview历史Alibaba40*46.9%69.7%10.228.9%88.8%3.7%闭源2026-04-20
55GPT-5.4 mini (xhigh)历史OpenAI39.981,17121.4%59.2%49.9%69.3%-18.6826.6%87.5%10%闭源2026-03-17
56Grok Build 0.1 0616历史SpaceXAI39.81,211.3910.9%52.1%50.2%64.7%7.4336%89.5%9.1%闭源2026-06-16
57Qwen3.6 PlusAlibaba39.561,135.3916.5%61.4%40.7%69.7%2.6525.7%88.2%2.9%闭源2026-04-02
58Gemini 3 Pro Preview (high)历史Google39.55*56.1%70.7%15.837.2%90.8%9.1%闭源2025-11-18
59GLM-5历史Z AI39.5*46.2%63.3%227.2%82%2%开放2026-02-11
60GPT-5.4 (low)历史OpenAI39.14*50.3%67.3%4.7228.9%87.1%7.4%闭源2026-03-05
61Qwen3.7 PlusAlibaba38.98936.2617.9%61%45.5%65%2.3733.4%90%9.1%闭源2026-06-01
62JT-4.1 Flash 236B A21BChina Mobile38.831,264.5428%59.6%38.2%58.7%-10.1216.1%84.5%0%闭源2026-07-09
63GPT-5.4 nano (xhigh)历史OpenAI38.241,099.8921%60.7%46.9%66%-29.5526.5%81.7%9.3%闭源2026-03-17
64MiniMax-M2.7历史MiniMax38.131,157.918.9%55.4%47%68.7%0.6828.1%87.4%0.6%开放2026-03-18
65GLM-5-Turbo历史Z AI38.06*43.6%60.7%-15.0825.4%84.7%0.3%闭源2026-03-15
66GPT-5.6 Luna (medium)OpenAI38.051,272.3715.3%53.2%45.8%66%-13.8224.5%85.9%4.9%闭源2026-07-09
67GPT-5.2 (medium)历史OpenAI37.95*46.2%63.3%0.0224.9%86.4%7.9%闭源2025-12-11
68Claude Opus 4.6 (high)历史Anthropic37.79*45.7%58.3%3.4718.6%84%2.8%闭源2026-02-05
69Gemini 3 Flash历史Google37.76*17.5%50.6%66.3%11.5734.7%89.8%8.6%闭源2025-12-17
70Nemotron 3 UltraNVIDIA37.761,163.813.8%53.9%39.9%67%-0.8326.6%86.7%3.1%开放2026-06-04
71Grok 4.3 (high)历史SpaceXAI37.581,084.912.2%39.7%47.3%64.3%18.3235%90.1%8%闭源2026-04-30
72DeepSeek V4 Flash (high)DeepSeek37.36*42%62.7%-22.2827.8%86.7%3.4%开放2026-04-24
73MiMo-V2.5Xiaomi37.21,145.016.6%63.7%43.1%62.7%-9.3325.2%84.9%3.7%开放2026-04-22
74Qwen3.6 27BAlibaba37.051,139.7215.3%60.7%39.8%68.7%-19.7821.6%84.2%1.1%开放2026-04-22
75Grok 4.20 0309 v2历史SpaceXAI37*45.6%58%15.3532.2%91.1%6.6%闭源2026-04-07
76GPT-5.1 (high)历史OpenAI36.87987.4514%52.4%43.3%75%5.5526.5%87.3%4.9%闭源2025-11-13
77Grok 4.20 0309历史SpaceXAI36.5*44.7%59%13.3730%88.5%6%闭源2026-03-10
78Claude 4.5 Sonnet历史Anthropic36.421,052.4619%55.8%44.7%65.7%0.3317.3%83.4%1.1%闭源2025-09-29
79MiMo-V2-Omni-0327Xiaomi36.39*39.5%63.7%-14.0220.4%85.5%0.9%闭源2026-03-27
80GPT-5 Codex (high)历史OpenAI36.12*40.9%69%-6.7725.6%83.7%5.1%闭源2025-09-23
81Grok 4.3 (medium)SpaceXAI36*44.6%65%16.9828.1%89%4.9%闭源2026-04-30
82Claude Sonnet 4.6 (Non-reasoning)历史Anthropic35.89*46.9%57.7%-2.9313.2%79.9%0.9%闭源2026-02-17
83Grok 4.3 (low)SpaceXAI35.44*41.9%64%13.6717.3%84.3%0.6%闭源2026-04-30
84GPT-5.5 (Non-reasoning)历史OpenAI35.381,118.9713.8%61%47.3%56.3%-5.0512.6%76.8%1.4%闭源2026-04-23
85Kimi K2.5历史Kimi35.371,008.5314.2%45.7%49%65.3%-8.1229.4%87.9%3.1%开放2026-01-27
86GLM-5.1历史Z AI35.37*36.1%44.3%-21.425.6%83.9%0%开放2026-04-07
87MiMo-V2-OmniXiaomi34.99*36.7%66.7%-17.4319.9%82.8%1.1%闭源2026-03-19
88Gemini 3.5 Flash (minimal)Google34.91*48.8%53.3%1.323.1%82.8%1.4%闭源2026-05-19
89GPT-5.1 Codex (high)历史OpenAI34.73*40.2%67.3%-6.0223.4%86%5.7%闭源2025-11-13
90Claude Opus 4.5历史Anthropic34.71*47%65.3%-3.9512.9%81%0.3%闭源2025-11-24
91GPT-5 (high)历史OpenAI34.711,074.819.6%35.2%42.9%75.6%-8.0826.5%85.4%5.7%闭源2025-08-07
92Kimi K2.6Kimi34.58*39.5%57.7%-9.9718.2%78.8%1.4%开放2026-04-20
93GLM 5V Turbo历史Z AI34.49*43.5%61%-18.9815.8%80.9%0.6%闭源2026-04-01
94Claude Sonnet 4.6 (Non-reasoning, Low Effort)Anthropic34.26*44.1%58.7%-1.910.8%79.7%0.9%闭源2026-02-17
95GLM-5.2Z AI34.051,387.5715.7%51.7%36.1%37.3%-6.48.4%68.6%3.1%开放2026-06-16
96GPT-5.6 Terra (Non-reasoning)OpenAI33.971,238.7213.4%56.2%44.6%50%-23.4711%74.6%2%闭源2026-07-09
97Qwen3.5 27B历史Alibaba33.78*39.5%67.3%-42.0222.2%85.8%0.9%开放2026-02-24
98GPT-5 (medium)历史OpenAI33.74*41.1%72.8%-10.0823.5%84.2%0%闭源2025-08-07
99GLM-4.7历史Z AI33.71,165.39.7%45.3%45.1%64%-34.625.1%85.9%1.7%开放2025-12-22
100KAT-Coder-Pro V2KwaiKAT33.69908.185.8%70%38.3%66%-21.616%85.5%0%闭源2026-03-27
101Qwen3.5 397B A17BAlibaba33.68961.9113.4%51.3%42%65.7%-29.7827.3%89.3%1.7%开放2026-02-16
102MiniMax-M2.5历史MiniMax33.65*42.6%66%-39.719.1%84.8%1.1%开放2026-02-12
103Hy3-previewTencent33.58*41.2%54.7%-34.6325.5%86.7%4.6%开放2026-04-23
104GPT-5.5 Instant (May 2026)历史OpenAI33.52*50.3%55.7%10.3320.3%84.6%2.6%闭源2026-05-05
105LongCat 2.0LongCat33.481,021.8412.8%50.2%35.4%58%-22.6332.1%78%2.6%开放2026-06-29
106Grok 4历史SpaceXAI33.28*45.7%68%3.7523.9%87.7%2%闭源2025-07-10
107GPT-5.6 Luna (low)OpenAI33.261,150.0812%43.4%45.6%59.3%-15.218.8%83.5%2.6%闭源2026-07-09
108MiMo-V2-Flash (Feb 2026)Xiaomi33.22*38.3%64.3%-18.4720%83.5%2.9%开放2025-12-16
109Gemini 3 Pro Preview (low)历史Google33.07*49.9%67.3%2.1527.6%88.7%0%闭源2025-11-18
110Kimi K2 Thinking历史Kimi32.7*42.4%66.3%-20.4822.3%83.8%2.6%开放2025-11-06
111GLM-5历史Z AI32.41*38.3%37%-11.237.2%66.6%0%开放2026-02-11
112Qwen3.5 122B A10BAlibaba32.28978.0213.6%47.6%42%66.7%-39.5823.4%85.7%0.6%开放2026-02-24
113DeepSeek V3.2历史DeepSeek32.04865.4818.8%46.8%38.9%65%-20.8822.2%84%2.9%开放2025-12-01
114Qwen3.5 397B A17BAlibaba31.98*41.1%58%-36.0818.8%86.1%0.9%开放2026-02-16
115Qwen3 Max Thinking历史Alibaba31.75*43.1%66%-34.4226.2%86.1%1.7%闭源2026-01-26
116Qwen3.6 35B A3BAlibaba31.651,048.678.7%44.9%35.8%63.7%-21.4320.2%84.1%0.3%开放2026-04-16
117MiniMax-M2.1历史MiniMax31.36*40.7%59%-32.8322.2%83%0.3%开放2025-12-23
118DeepSeek V4 ProDeepSeek31.22*42.4%45%-30.087.7%71.7%0.9%开放2026-04-24
119MiMo-V2-Flash历史Xiaomi31.2*39.4%63%-43.4521.1%84.6%4.3%开放2025-12-16
120GPT-5 (low)历史OpenAI31.15*39.1%58.7%-10.0218.4%80.8%1.1%闭源2025-08-07
121GPT-5 mini (medium)历史OpenAI30.92*41%66%-10.7814.6%80.3%1.4%闭源2025-08-07
122Qwen3.5 Omni PlusAlibaba30.64*40.5%52.7%-12.313.9%82.6%0.6%闭源2026-03-30
123GPT-5.1 Codex mini (high)历史OpenAI30.63*42.6%62.7%-16.5216.9%81.3%0%闭源2025-11-13
124Grok 4.1 Fast历史SpaceXAI30.62*44.2%68%-28.7317.6%85.3%2.9%闭源2025-11-19
125Ring-2.6-1TInclusionAI30.57918.7814.2%43.1%42.4%64.3%-37.7518.3%85.7%3.7%开放2026-05-08
126Qwen3.6 27BAlibaba30.461,112.728.2%51.3%37.3%55%-52.7813.6%82.9%0.9%开放2026-04-22
127DeepSeek V3.1 Terminus历史DeepSeek30.44889.2815.9%44.9%40.6%65%-24.3215.2%79.2%1.7%开放2025-09-22
128o3OpenAI30.4*41%69.3%-15.2720%82.7%1.1%闭源2025-04-16
129Step 3.7 FlashStepFun30.271,017.0411.3%39.3%40%63.7%-37.5519.9%80.9%2.3%开放2026-05-29
130GPT-5.4 nano历史OpenAI30.16*38.4%57.3%-18.8214.7%76.1%5.1%闭源2026-03-17
131Mistral Medium 3.5Mistral29.95928.9514.4%50.6%39.6%61%-36.3212.8%74.8%0%开放2026-04-29
132GPT-5.4 mini (medium)历史OpenAI29.81*44.2%61.3%-20.1217.1%82.3%2.9%闭源2026-03-17
133Claude 4.5 HaikuAnthropic29.589079.1%44.2%43.3%70.3%-4.229.7%67.2%0%闭源2025-10-15
134Kimi K2.5历史Kimi29.4*39.6%59%-1412.3%78.9%0.6%开放2026-01-27
135Gemma 4 31BGoogle29.35804.0315.1%43.4%43.4%62%-45.4222.7%85.7%1.4%开放2026-04-02
136Qwen3.5 27B历史Alibaba29.31*36.7%55.7%-46.7513.2%84.2%0.3%开放2026-02-24
137Claude 4.5 Sonnet历史Anthropic29.28*42.8%51.3%-9.27.1%72.7%0%闭源2025-09-29
138Qwen3.5 35B A3B历史Alibaba29.26*37.7%62.7%-46.3819.7%84.5%0.9%开放2026-02-24
139Claude 4 Sonnet历史Anthropic28.92860.5513.8%36.3%40%64.7%0.189.6%77.7%0.3%闭源2025-05-22
140GPT-5.5 Instant (June 2026)OpenAI28.86718.5611.5%34.8%48.6%64%3.4818.6%82.3%0%闭源2026-06-25
141GLM-4.6历史Z AI28.7933.5710.5%49.4%38.4%54.3%-41.7313.3%78%1.1%开放2025-09-30
142DeepSeek V4 FlashDeepSeek28.65*37.3%33.3%-44.187%71.6%0.3%开放2026-04-24
143JT-35B-FlashChina Mobile28.36*29.1%55.3%-23.256.1%82.9%0%闭源2026-05-14
144MiniMax-M2历史MiniMax28.31*36.1%61%-46.9312.5%77.7%0.9%开放2025-10-26
145KAT-Coder-Pro V1KwaiKAT28.26*898.5936.6%74%-37.4333.4%76.4%0%闭源2025-11-11
146MiMo-V2.5-ProXiaomi27.86*39.1%35%-38.313.3%76.2%1.1%开放2026-04-22
147GPT-5.4 (Non-reasoning)历史OpenAI27.68*47.1%47.3%-15.8810.6%74.8%0.6%闭源2026-03-05
148Qwen3.5 122B A10BAlibaba27.6879.919.1%47.2%35.6%56%-53.914.8%82.7%0.9%开放2026-02-24
149Gemini 3 Flash历史Google27.37*49.9%48%-3.6214.1%81.2%1.4%闭源2025-12-17
150Grok 4 Fast历史SpaceXAI27.37*44.2%64.7%-28.417%84.7%2.9%闭源2025-09-19
151Claude 3.7 Sonnet历史Anthropic27.06*40.3%60.7%0.3210.3%77.2%0.9%闭源2025-02-24
152GPT-5.6 Luna (Non-reasoning)OpenAI26.561,068.169.1%39%39.9%36.3%-25.126.7%64.5%0.3%闭源2026-07-09
153GLM-4.7历史Z AI26.56*35.4%36.3%-46.286.1%66.4%0%开放2025-12-22
154Hy3-previewTencent26.1*39.4%33.7%-36.46.3%73.2%0.3%开放2026-04-23
155Ling-2.6-1TInclusionAI26.05*37%34.7%-518.2%75.2%0.3%开放2026-04-23
156GPT-5.2历史OpenAI26.02*12%40.4%38%-13.727.3%71.2%0.6%闭源2025-12-11
157Step 3.5 Flash 2603StepFun26*38.5%54.3%-43.6722.6%82.6%2.3%闭源2026-04-02
158Doubao Seed CodeByteDance Seed25.98*40.7%65.3%-34.4313.3%76.4%0.3%闭源2025-11-11
159Gemini 2.5 ProGoogle25.82665.399.3%28.5%42.8%66%-14.321.1%84.4%2.6%闭源2025-06-05
160Gemma 4 26B A4BGoogle25.69761.3311.8%39%40%55.7%-48.0718.3%79.2%0%开放2026-04-02
161o4-mini (high)历史OpenAI25.55*46.5%55%-35.7517.5%78.4%0.6%闭源2025-04-16
162Step 3.5 Flash历史StepFun25.5*40.4%43%-40.919.1%83.1%2.5%开放2026-02-02
163Claude 4 Sonnet历史Anthropic25.49*37.3%44.3%-9.234%68.3%1.1%闭源2025-05-22
164DeepSeek V3.2 Exp历史DeepSeek25.45*37.7%69%-29.9313.8%79.7%1.4%开放2025-09-29
165NVIDIA Nemotron 3 SuperNVIDIA25.41693.2510.1%38.6%36%60%-42.0719.2%80%3.1%开放2026-03-11
166GPT-5 mini (high)历史OpenAI25.31936.7114.6%3.7%39.2%68%-17.1819.7%82.8%0%闭源2025-08-07
167Gemini 3.1 Flash-LiteGoogle25.046428.7%31.1%41.9%65.3%-15.5216.2%82.2%1.1%闭源2026-03-03
168Qwen3 Max Thinking (Preview)历史Alibaba25.03*38.7%57.7%-37.8312%77.6%0%闭源2025-11-03
169Grok 4.3 (Non-reasoning)SpaceXAI24.751,095.967.6%34.1%37.4%24.7%-32.36.5%65.8%0%闭源2026-04-30
170K-EXAONELG AI Research24.7*35.6%55.7%-57.9313.1%78.3%1.1%开放2025-12-31
171MiMo-V2-FlashXiaomi24.67833.053.1%61.8%25.9%31.3%-48.488%65.6%0%开放2025-12-16
172DeepSeek V3.2历史DeepSeek24.66*38.7%39%-46.7310.5%75.1%0.9%开放2025-12-01
173Trinity Large ThinkingArcee AI24.47*553.8436.1%33%-44.1814.7%75.2%0.9%开放2026-04-01
174Qwen3.6 35B A3BAlibaba24.161,020.435.2%41.6%1.3%56.7%-59.8312.5%81.7%0%开放2026-04-16
175Qwen3 Max历史Alibaba24.01*38.3%46.7%-43.1211.1%76.4%0%闭源2025-09-23
176Qwen3.5 35B A3BAlibaba23.99796.954.9%40.8%29.3%55.3%-6212.8%81.9%0.6%开放2026-02-24
177gpt-oss-120b (high)OpenAI23.83799.3312%26.2%38.9%50.7%-50.0518.5%78.2%1.1%开放2025-08-05
178Gemini 2.5 Flash (Sep)历史Google23.8*40.5%64.3%-34.7212.7%79.3%0.3%闭源2025-09-25
179Claude 4.5 HaikuAnthropic23.71*34.4%43.7%-7.674.3%64.6%0%闭源2025-10-15
180Kimi K2 0905历史Kimi23.54*30.7%52.3%-26.476.3%76.7%0%开放2025-09-05
181Claude 3.7 Sonnet历史Anthropic23.5*37.6%48.3%-9.234.8%65.6%0%闭源2025-02-24
182o1历史OpenAI23.44*35.8%59.3%-10.557.7%74.7%0.3%闭源2024-12-05
183GLM-4.6历史Z AI22.98*33.1%26.3%-31.65.2%63.2%0%开放2025-09-30
184EXAONE 4.5 33BLG AI Research22.96*28%49.3%-51.4811.6%79.4%0.3%开放2026-04-09
185GLM-4.7-Flash历史Z AI22.89*33.7%35%-59.337.1%58.1%0.3%开放2026-01-19
186Command A+Cohere22.51714.25.8%22.8%37.8%46%-3.9811.4%76.1%0.3%开放2026-05-20
187Grok 3 mini Reasoning (high)历史SpaceXAI22.5*40.6%50.3%-6.0511.1%79.1%0.6%闭源2025-02-19
188Grok 4.20 0309历史SpaceXAI22.48*32.2%18%-42.6822.5%78.5%0%闭源2026-03-10
189DeepSeek V3.2 Speciale历史DeepSeek22.24*44%59.3%-16.4826.1%87.1%7.4%开放2025-12-01
190Gemma 4 12BGoogle22*38.2%55.3%-51.8814.8%75.3%0%开放2026-06-03
191ERNIE 5.0 Thinking PreviewBaidu21.92*37.5%6.7%-44.1312.7%77.7%1.4%闭源2025-11-13
192Grok 4.20 0309 v2历史SpaceXAI21.84*32.8%17.3%-44.624.2%77.6%0.3%闭源2026-04-07
193Nova 2.0 Pro Preview (medium)Amazon21.77676.227.6%29.6%42.7%54.3%-48.058.9%78.5%0%闭源2025-11-27
194Gemma 4 31BGoogle21.77744.228.5%29.2%41.1%36%-51.3511.5%76.3%0%开放2026-04-02
195Grok Code Fast 1历史SpaceXAI21.61*36.2%48.3%-367.5%72.7%0%闭源2025-08-28
196Qwen3.5 9BAlibaba21.44635.588.2%29.2%27.5%59%-52.5213.3%80.6%0.3%开放2026-03-02
197Mercury 2Inception21.43697.619.9%27.3%38.7%36.3%-52.2815.5%77%0.8%闭源2026-02-20
198DeepSeek V3.1 Terminus历史DeepSeek21.41*32.1%43.3%-41.888.4%75.1%0%开放2025-09-22
199DeepSeek V3.2 Exp历史DeepSeek21.33*39.9%43%-47.088.6%73.8%1.4%开放2025-09-29
200Nemotron Cascade 2 30B A3BNVIDIA21.25*34.8%34%-52.4711.4%75.8%0.6%开放2026-03-19
201Apriel-v1.5-15B-Thinker历史ServiceNow21.23*34.8%20%-55.7812%71.3%1.1%开放2025-09-30
202Qwen3 Coder NextAlibaba21.11711.265.2%38.2%32.3%40%-60.729.3%73.7%0%开放2026-02-03
203DeepSeek V3.1历史DeepSeek21.05*36.7%45%-41.076.3%73.5%0%开放2025-08-21
204Nova 2.0 Omni (medium)Amazon20.95*36.2%53.7%-57.836.8%76%0%闭源2025-11-26
205DeepSeek V3.1历史DeepSeek20.67*39.1%53.3%-28.3713%77.9%2%开放2025-08-21
206Qwen3 VL 235B A22B历史Alibaba20.6*39.9%58.7%-44.6210.1%77.2%0%开放2025-09-23
207Apriel-v1.6-15B-ThinkerServiceNow20.54*37.3%50.3%-58.539.8%73.3%0.3%开放2025-11-25
208GPT-5.1历史OpenAI20.4*36.5%44%-34.955.2%64.3%0%闭源2025-11-13
209Qwen3.5 9BAlibaba20.32*4.1%21.3%27.7%38%-70.758.6%78.6%0.6%开放2026-03-02
210Gemma 4 26B A4BGoogle20.1*37.3%39.7%-62.0310.7%71.4%0%开放2026-04-02
211Qwen3.5 4BAlibaba20.09*8.2%25.8%16.1%55.7%-56.777.8%77.1%0%开放2026-03-02
212DeepSeek R1 0528历史DeepSeek20.08*40.3%54.7%-27.0814.9%81.3%1.4%开放2025-05-28
213Gemini 2.5 Flash历史Google20.06*39.4%61.7%-27.8511.1%79%1.1%闭源2025-05-20
214GPT-5 nano (high)历史OpenAI19.87*36.6%41.7%-27.758.2%67.6%0%闭源2025-08-07
215North Mini CodeCohere19.8533.976.4%35.6%38.2%32.3%-49.0710%75.7%0.3%开放2026-06-09
216Qwen3 235B A22B 2507历史Alibaba19.61540.617.8%12%42.4%67%-45.5515%79%0%开放2025-07-25
217Nova 2.0 Pro Preview (low)Amazon19.61645.199.1%19.5%38.7%61.7%-45.775.2%75.1%0%闭源2025-11-27
218Mistral Small 4Mistral19.56587.995.2%21%38%44.7%-29.99.5%76.9%0.3%开放2026-03-16
219GLM-4.5历史Z AI19.49*34.8%48.3%-25.4512.2%78.2%0%开放2025-07-28
220Kimi K2历史Kimi19.4*34.5%51%-27.527%76.6%0%开放2025-07-11
221GPT-4.1历史OpenAI19.36*38.1%61%-36.184.6%66.6%0%闭源2025-04-14
222Qwen3 Max (Preview)历史Alibaba19.18*37%39.7%-40.589.3%76.4%0.9%闭源2025-09-05
223Devstral 2Mistral19.16739.0510.3%30.3%33.1%30%-46.173.6%59.4%0%开放2025-12-09
224GPT-5 nano (medium)历史OpenAI19*33.8%40%-24.887.6%67%0%闭源2025-08-07
225Nova 2.0 Lite (medium)Amazon19*36.8%58.3%-55.628.6%76.8%0%闭源2025-10-29
226Qwen3.5 Omni FlashAlibaba18.99*25.5%44%-65.637.1%74.2%0%闭源2026-03-30
227Gemini 2.5 Flash (Sep)历史Google18.83*37.5%56.7%-38.127.8%76.6%0%闭源2025-09-25
228JT-MINIChina Mobile18.53*27.2%11.7%-64.586.6%67.6%0%闭源2026-04-15
229DeepSeek R1 (Jan)历史DeepSeek18.51525.056.4%19.1%35.7%52.3%-31.329.3%70.8%0.6%开放2025-01-20
230Grok 3历史SpaceXAI18.35*36.8%54.7%-32.335.1%69.3%0%闭源2025-02-19
231Seed-OSS-36B-Instruct历史ByteDance Seed18.34*36.5%57.7%-50.859.1%72.6%0%开放2025-08-20
232Nova 2.0 Lite (high)Amazon18.17585.556.8%16.1%36.9%55.3%-54.1210.9%81.1%0.3%闭源2025-10-29
233Qwen3 235B 2507历史Alibaba18.16*36%31.2%-42.110.6%75.3%0%开放2025-07-21
234Qwen3 Coder 480B历史Alibaba17.98*35.9%42.3%-22.24.4%61.8%0%开放2025-07-22
235Qwen3 VL 32B历史Alibaba17.93*28.5%55.3%-50.29.6%73.3%0%开放2025-10-21
236Magistral Medium 1.2Mistral17.87402.935.6%12.4%39.2%51.3%-26.239.6%73.9%0.3%闭源2025-09-18
237Nova 2.0 Lite (low)Amazon17.81*33.3%52%-51.074.2%69.8%0%闭源2025-10-29
238HyperNova 60B 2605Multiverse Computing17.8653.655.4%18.4%33%31.7%-56.4315.1%73.3%0.9%开放2026-05-26
239MiniMax M1 80k历史MiniMax17.67*37.4%54.3%-47.428.2%69.7%0%开放2025-06-17
240GPT-5.4 nano历史OpenAI17.61*35.2%24.7%-38.584.2%55.8%0%闭源2026-03-17
241Devstral Small 2Mistral17.44728.839.7%29.6%28.8%24%-56.733.4%53.2%0%开放2025-12-09
242K2 Think V2MBZUAI Institute of Foundation Models17.26369.865.4%15%33%52.7%-33.929.5%71.3%0%开放2025-12-15
243LongCat Flash LiteLongCat17.22*28.4%25.7%-70.16%63.6%0%开放2026-01-28
244GPT-5 (minimal)历史OpenAI17.18*38.8%25%-33.955.4%67.3%0%闭源2025-08-07
245HyperCLOVA X SEED Think (32B)Naver17.02*28.4%11.7%-52.875.5%61.5%0%开放2025-12-26
246Grok 4.1 Fast历史SpaceXAI16.88*29.6%22%-50.955%63.7%0%闭源2025-11-19
247GLM-4.6V历史Z AI16.75*30.4%40.3%-26.488.9%71.9%0%开放2025-12-08
248K-EXAONELG AI Research16.74*27%47%-61.175.4%69.5%0%开放2025-12-31
249Qwen3 Next 80B A3BAlibaba16.72367.356.2%6.7%38.8%60.3%-50.5711.7%75.9%0%开放2025-09-11
250GPT-5.4 mini历史OpenAI16.62*784.3139.6%30.3%-45.855.7%60.6%0%闭源2026-03-17
251Nova 2.0 Omni (low)Amazon16.56*34.3%51%-49.984%69.9%0%闭源2025-11-26
252GLM-4.5-Air历史Z AI16.52*30.6%43.7%-62.526.8%73.3%0%开放2025-07-28
253Grok 4 Fast历史SpaceXAI16.47*32.9%20%-49.875%60.6%0%闭源2025-09-19
254Mi:dm K 2.5 ProKorea Telecom16.42*33.2%9%-57.277.7%70.1%0%闭源2025-12-11
255Ring-1T历史InclusionAI16.16*36.7%45.7%-47.8710.2%77.4%0.6%开放2025-10-13
256Qwen3.5 4BAlibaba16*3.1%21.3%18.3%28.3%-74.727.5%71.2%0%开放2026-03-02
257Mistral Large 3Mistral15.93632.515.8%12%36.2%34.7%-39.434.1%68%0%开放2025-12-02
258o3-mini (high)历史OpenAI15.64468.115.2%4.5%39.8%39.3%-41.3312.3%77.3%0.3%闭源2025-01-31
259INTELLECT-3Prime Intellect15.61*39.1%32.3%-51.0212.1%76.1%0.3%开放2025-11-27
260GLM-4.7-Flash历史Z AI15.52*25.5%14.7%-62.234.9%45.2%0%开放2026-01-19
261DeepSeek V3 0324历史DeepSeek15.43314.484.5%13.9%35.8%41%-37.575.2%65.5%0%开放2025-03-25
262Solar Open 100BUpstage15.15*26.9%36%-54.19.2%65.7%0%开放2025-12-17
263Gemini 2.5 Flash-Lite (Sep)历史Google15.13*28.7%59%-53.086.6%70.9%0%闭源2025-09-08
264Nemotron 3 Nano Omni 30B A3B ReasoningNVIDIA14.93*27.8%35.7%-56.035.3%46.9%0%开放2026-04-29
265gpt-oss-120b (low)OpenAI14.91396.552.9%13.9%36%43.7%-50.155.2%67.2%0%开放2025-08-05
266gpt-oss-20b (high)OpenAI14.89559.037%13.9%34.4%30.7%-63.929.8%68.8%1.4%开放2025-08-05
267GPT-4.1 mini历史OpenAI14.77502.925.2%10.1%40.4%42.3%-50.134.6%66.4%0%闭源2025-04-14
268Mistral Small 3.1历史Mistral14.74590.157%26.2%26.5%19.7%-50.324.8%45.4%0%开放2025-03-17
269Mistral Medium 3.1历史Mistral14.68599.258.5%13.9%33.8%19.7%-45.474.4%58.8%0%闭源2025-08-12
270Qwen3 30B A3B 2507历史Alibaba14.44307.725.4%1.5%33.3%59%-56.879.8%70.7%0.3%开放2025-07-30
271Nova 2.0 Pro PreviewAmazon14.36554.736.8%17.2%28.1%28.3%-48.154%63.6%0%闭源2025-11-27
272gpt-oss-20b (low)OpenAI14.34*34%31%-60.135.1%61.1%0%开放2025-08-05
273Qwen3 VL 235B A22B历史Alibaba14.31*35.9%31.7%-50.676.3%71.2%0%开放2025-09-23
274Llama 4 MaverickMeta14.27-16.153.9%7.9%33.1%46%-41.774.8%67.1%0%开放2025-04-05
275GPT-5 mini (minimal)历史OpenAI14.25*36.9%35.7%-54.125%68.7%0%闭源2025-08-07
276K2-V2 (high)MBZUAI Institute of Foundation Models14.19*28.6%33.3%-56.79.8%68.1%0%开放2025-12-05
277NVIDIA Nemotron 3 NanoNVIDIA14.18483.846%6.7%29.6%33.7%-51.6310.2%75.7%0.9%开放2025-12-15
278DeepSeek V3 (Dec)历史DeepSeek14.16217.094.7%16.9%35.4%29%-41.353.6%55.7%0%开放2024-12-26
279Gemini 2.5 Flash历史Google14.14*29.1%45.9%-42.025.1%68.3%1.4%闭源2025-05-20
280Solar Pro 3Upstage14.14492.558%12%24.7%27%-53.7810.1%72.4%0%闭源2026-04-06
281Ling 2.6 FlashInclusionAI14.05544.692.9%24.3%27.1%25%-65.686.2%59.3%0%开放2026-04-21
282Qwen3 Next 80B A3BAlibaba13.72*30.7%51.3%-59.17.3%73.8%0%开放2025-09-11
283Tri-21B-think PreviewTrillion Labs13.62*17.8%14.7%-55.285.7%53.8%0%开放2026-02-10
284Qwen3 Coder 30B A3B历史Alibaba13.61*27.8%29%-50.854%51.6%0%开放2025-07-31
285DiffusionGemma 26B A4BGoogle13.45547.266.6%12.4%34.3%14.3%-59.0310.2%66.9%0.3%开放2026-06-10
286Qwen3 235B历史Alibaba13.43*39.9%0%-44.5811.7%70%0%开放2025-04-28
287Qwen3 VL 30B A3B历史Alibaba13.34*28.8%40.7%-57.328.7%72%0%开放2025-10-03
288Gemma 4 12B (Non-reasoning)Google13.19*29.7%30.7%-52.736.2%66.1%0%开放2026-06-03
289Gemini 2.5 Flash-Lite (Sep)历史Google13.1*28.5%48%-42.674.6%65.1%0%闭源2025-09-25
290Motif-2-12.7BMotif Technologies12.79*28.2%13%-61.188.2%69.5%0%闭源2025-12-04
291Ling-1T历史InclusionAI12.75*35.2%34.7%-55.487.2%71.9%0.6%开放2025-10-08
292Nova PremierAmazon12.73*27.9%30%-36.274.7%56.9%0%闭源2025-04-30
293Magistral Medium 1历史Mistral12.51*29.7%0%-26.429.5%67.9%0.3%闭源2025-06-10
294Gemma 4 E4BGoogle12.5*24.4%30.7%-20.053.7%57.6%0.6%开放2026-04-03
295Mistral Medium 3历史Mistral12.49*33.1%28%-31.474.3%57.8%0%闭源2025-05-07
296K2-V2 (medium)MBZUAI Institute of Foundation Models12.43*25.2%28%-49.974.4%59.8%0%开放2025-12-05
297Llama Nemotron Super 49B v1.5NVIDIA12.42*34.8%34%-45.826.8%74.8%0%开放2025-07-25
298Devstral Medium历史Mistral12.4*29.4%28.7%-31.623.8%49.2%0%闭源2025-07-10
299Mistral Small 4Mistral12.37*28.1%21.3%-48.373.7%57.1%0.3%开放2026-03-16
300Tri-21B-ThinkTrillion Labs12.36*17.4%11%-63.36.1%60.1%0.3%开放2026-02-10
301Gemini 2.0 Flash历史Google12.27*33.3%28.3%-42.555.3%62.3%0%闭源2025-02-05
302Claude 3.5 Haiku历史Anthropic12.26444.215.8%10.1%27.4%23.3%-23.183.5%40.8%0%闭源2024-10-22
303Llama 3.3 Nemotron Super 49B历史NVIDIA12.25*28.2%17%-58.336.5%64.3%0%开放2025-03-18
304MiniCPM5-1BOpenBMB11.96*4.4%3.7%-16.626.5%27.8%0%开放2026-05-25
305Qwen3 4B 2507历史Alibaba11.96*25.6%37.7%-60.825.9%66.7%0%开放2025-08-06
306Sarvam 105B (high)Sarvam11.94*26.4%0%-59.5310.1%73.8%0%开放2026-03-06
307Nova 2.0 LiteAmazon11.83*24%17.7%-59.13%60.3%0%闭源2025-10-29
308Devstral Small (May)历史Mistral11.82*24.5%26.7%-56.984%43.4%0%开放2025-05-21
309MiniCPM5-1BOpenBMB11.74*1.4%4.7%-0.784.6%26.9%0%开放2026-05-25
310Qwen3 32B历史Alibaba11.55274.985.4%5.2%35.4%0%-50.538.3%66.8%0.3%开放2025-04-28
311Gemini 2.5 Flash-Lite历史Google11.41*19.3%51.3%-44.636.4%62.5%0%闭源2025-06-17
312Magistral Small 1.2Mistral11.29248.364.3%4.5%35.2%16.3%-64.986.1%66.3%0.3%开放2025-09-17
313GPT-4o (Nov)历史OpenAI11.18*33.3%0%-10.753.3%54.3%0%闭源2024-11-20
314Nanbeige4.1-3BNanbeige11.09-141.110%1.1%26.6%0%-41.9310%84.9%0%开放2026-02-11
315Qwen3 VL 32B历史Alibaba11.06*30.1%31.3%-62.826.3%67.1%0%开放2025-10-21
316Ministral 3 14BMistral11.05475.866.6%9.7%23.6%22%-66.824.6%57.2%0%开放2025-12-02
317GLM-4.6V历史Z AI10.98*27.2%12.3%-37.873.7%56.6%0%开放2025-12-08
318Qwen3 235B历史Alibaba10.86*29.9%0%-52.974.7%61.3%0%开放2025-04-28
319Magistral Small 1历史Mistral10.7*24.1%0%-43.377.2%64.1%0%开放2025-06-10
320EXAONE 4.0 32BLG AI Research10.59*34.4%14%-60.810.5%73.9%0%开放2025-07-15
321Qwen3 VL 8B历史Alibaba10.58*21.9%31%-52.423.3%57.9%0.3%开放2025-10-14
322Mistral Small 3.2历史Mistral10.5657.176%5.6%26.4%17.3%-50.424.3%50.5%0%开放2025-06-20
323Nova 2.0 OmniAmazon10.53*27.9%22.3%-64.53.9%55.5%0%闭源2025-11-26
324Qwen3 14B历史Alibaba10.38222.455.4%4.9%31.6%0%-49.234.3%60.4%0%开放2025-04-28
325DeepSeek R1 0528 Qwen3 8B历史DeepSeek10.37*20.4%13%-64.125.6%61.2%0%开放2025-05-29
326Llama 4 ScoutMeta10.0489.683.3%3.7%17%25.8%-52.374.3%58.7%0%开放2025-04-05
327Qwen3 VL 30B A3B历史Alibaba10.02*30.8%23.7%-636.4%69.5%0%开放2025-10-03
328Hermes 4 70BNous Research9.97*34.1%6.7%-49.927.9%69.9%0%开放2025-08-27
329DeepSeek R1 Distill Llama 70B历史DeepSeek9.93*31.3%11%-46.456.1%40.2%0%开放2025-01-20
330Falcon-H1R-7BTII UAE9.79*24.9%8.7%-62.0810.8%66.1%0.3%开放2026-01-04
331Ling-flash-2.0历史InclusionAI9.74*28.9%15%-63.376.3%65.7%0%开放2025-09-17
332Qwen3 Omni 30B A3BAlibaba9.63*30.6%0%-60.727.3%72.6%0%开放2025-09-22
333GPT-4o (Aug)历史OpenAI9.58*33.1%35%-21.722.9%52.1%0%闭源2024-08-06
334GPT-4.1 nano历史OpenAI9.5841.43.5%3.7%25.9%17%-56.423.9%51.2%0%闭源2025-04-14
335Qwen2.5 72B历史Alibaba9.57*26.7%20.3%-52.224.2%49.1%0%开放2024-09-19
336Step3 VL 10BStepFun9.47*31.1%0%-5910.2%69%0%开放2026-01-20
337Llama 3.3 70BMeta9.479.941%4.9%26%15%-51.924%49.8%0%开放2024-12-06
338Qwen3 30B历史Alibaba9.31*28.5%0%-51.26.6%61.6%0%开放2025-04-28
339Gemma 4 E2BGoogle9.26*20.9%15%-23.984.8%43.3%0%开放2026-04-02
340Devstral Small历史Mistral9.26*24.3%17%-51.153.7%41.4%0%开放2025-07-10
341Mistral Large 2 (Nov)历史Mistral9.15*29.2%5.3%-34.054%48.6%0%开放2024-11-18
342GLM-4.5V历史Z AI9.15*22.1%0%-43.575.9%68.4%0%开放2025-08-11
343Llama Nemotron UltraNVIDIA9.08*34.7%7.3%-45.528.1%72.8%0%开放2025-04-07
344Qwen3 30B A3B 2507历史Alibaba9.06*30.4%22.7%-65.926.8%65.9%0%开放2025-07-29
345ERNIE 4.5 300B A47BBaidu9.02*31.5%2.3%-35.883.5%81.1%0%开放2025-06-30
346Hermes 4 405BNous Research9*25.2%20.7%-36.0510.3%72.7%0.3%开放2025-08-27
347Solar Pro 2Upstage8.99*30.2%0%-56.577%68.7%0%闭源2025-07-09
348NVIDIA Nemotron Nano 12B v2 VLNVIDIA8.96*26.2%40%-64.425.3%57.2%0%开放2025-10-28
349Ministral 3 8BMistral8.96448.723.7%4.1%20.8%24%-67.834.3%47.1%0%开放2025-12-02
350Gemma 4 E4BGoogle8.91*3.9%18%-41.524.7%54.9%0.3%开放2026-04-03
351Granite 4.1 30BIBM8.86422.013.9%2.6%25.8%18.7%-67.784.2%48.1%0%开放2026-04-29
352NVIDIA Nemotron Nano 9B V2NVIDIA8.84*22%21%-42.874.6%57%0%开放2025-08-18
353Hermes 4 405BNous Research8.8*34.6%20%-334.2%53.6%0%开放2025-08-27
354NVIDIA Nemotron 3 Nano 4BNVIDIA8.73192.397%3.7%16.4%16.7%-71.534.8%51.3%0%开放2026-03-16
355Llama Nemotron Super 49B v1.5NVIDIA8.68*23.8%22%-45.634.3%48.1%0%开放2025-07-25
356K2-V2 (low)MBZUAI Institute of Foundation Models8.55*22.3%19%-48.073.9%54.1%0%开放2025-12-05
357Llama 3.1 405BMeta8.5*29.9%24.3%-17.34.2%51.5%0%开放2024-07-23
358Llama 3.3 Nemotron Super 49B历史NVIDIA8.47*22.9%11.3%-49.683.5%51.7%0%开放2025-03-18
359Qwen3 VL 8B历史Alibaba8.42*17.4%15.3%-52.052.9%42.7%0%开放2025-10-14
360LFM2.5-8B-A1BLiquid AI8.32*7.8%0%-33.276.9%51.3%0%开放2026-05-28
361Qwen3 8B历史Alibaba8.29200.344.5%2.2%22.6%0%-65.074.2%58.9%0%开放2025-04-28
362Ring-flash-2.0InclusionAI8.16*16.8%21%-58.678.9%72.5%0.3%开放2025-09-19
363Olmo 3.1 32B ThinkAllen Institute for AI8.09*29.3%0%-44.126%59.1%0%开放2025-12-12
364GPT-5 nano (minimal)历史OpenAI8*29.1%20%-63.984.1%42.8%0%闭源2025-08-07
365Qwen3 VL 4B历史Alibaba7.9*17.1%21.3%-69.354.4%49.4%0%开放2025-10-14
366Solar Pro 2Upstage7.77*24.8%0%-61.683.8%56.1%0%闭源2025-07-09
367Nova Pro历史Amazon7.67*20.8%19%-47.583.4%49.9%0%闭源2024-12-03
368Command ACohere7.67*28.1%18%-48.474.6%52.7%0%开放2025-03-13
369Llama 3.1 Nemotron 70BNVIDIA7.64*23.3%7%-41.054.6%46.5%0%开放2024-10-15
370Qwen3.5 2BAlibaba7.64198.862.5%3%2.8%23.7%-47.482.1%45.6%0%开放2026-03-02
371Llama 3.1 8B历史Meta7.61-123.731.6%1.5%13.2%15.7%-30.335.1%25.9%0%开放2024-07-23
372Gemma 3 27B历史Google7.42-144.460.8%4.5%21.2%5.7%-65.884.7%42.8%0%开放2025-03-12
373NVIDIA Nemotron 3 NanoNVIDIA7.39*-94.3423%6.7%-69.224.6%39.9%0%开放2025-12-15
374NVIDIA Nemotron Nano 9B V2NVIDIA7.38*20.9%22.7%-57.14%55.7%0%开放2025-08-18
375Qwen3 4B 2507历史Alibaba7.12*18.1%7.3%-59.634.7%51.7%0%开放2025-08-06
376Qwen3 14B历史Alibaba7.02*26.5%0%-67.384.2%47%0%开放2025-04-28
377GLM-4.5V历史Z AI7*18.8%0%-53.523.6%57.3%0%开放2025-08-11
378Nova Lite历史Amazon6.92*13.9%17.7%-40.424.6%43.3%0%闭源2024-12-03
379Gemini 2.5 Flash-Lite历史Google6.92*17.7%31.3%-573.7%47.4%0%闭源2025-06-17
380Hermes 4 70BNous Research6.9*27.7%2%-47.373.6%49.1%0%开放2025-08-27
381Ministral 3 3BMistral6.8272.84.7%0%14.4%11.7%-64.15.3%35.8%0%开放2025-12-02
382Qwen3 30B历史Alibaba6.8*26.4%0%-65.074.6%51.5%0%开放2025-04-28
383Llama 3.1 70B历史Meta6.75*26.7%6.3%-40.854.6%40.9%0%开放2024-07-23
384Granite 4.1 8BIBM6.67*3.1%3.4%21.8%12%-64.63.8%43.3%0%开放2026-04-29
385Sarvam 30B (high)Sarvam6.64*19.2%0%-72.057%63.3%0.3%开放2026-03-06
386Olmo 3.1 32B InstructAllen Institute for AI6.47*16.7%0%-51.084.9%53.9%0%开放2026-01-13
387Gemma 4 E2BGoogle6.41*20.4%15%-61.374.5%40.5%0%开放2026-04-02
388Olmo 3 32B Think历史Allen Institute for AI6.4*28.6%0%-58.45.9%61%0%开放2025-11-20
389Phi-4 MiniMicrosoft6.01-145.051%0.4%10.8%13.7%-61.024.2%33.1%0%开放2024-02-26
390EXAONE 4.0 32BLG AI Research6.01*25.2%8%-62.34.9%62.8%0%开放2025-07-15
391Qwen3.5 2BAlibaba5.62-103.622.3%0%7.2%13.7%-82.834.9%43.8%0%开放2026-03-02
392Gemma 3 12B历史Google5.51-143.640.8%0%17.4%6.7%-76.984.8%34.9%0%开放2025-03-12
393Qwen3.5 0.8BAlibaba5.48-111.260%0%0%5.3%-35.181.2%11.4%0%开放2026-03-02
394Jamba 1.7 LargeAI21 Labs5.3*18.8%17.3%-55.553.8%39%0%开放2025-07-07
395Granite 4.0 H SmallIBM5.24*20.9%9%-60.923.7%41.6%0%开放2025-09-22
396Qwen3 Omni 30B A3BAlibaba5.11*18.6%0%-69.185.1%62%0%开放2025-09-22
397Qwen3 8B历史Alibaba5.07*16.8%0%-74.12.8%45.2%0%开放2025-04-28
398LFM2 24B A2BLiquid AI4.95*10.9%0%-59.074.4%47.4%0%开放2026-02-25
399Phi-4Microsoft4.87*26%0%-56.724.1%57.5%0%开放2024-12-12
400Nova MicroAmazon4.74*9.4%9.7%-48.674.7%35.8%0%闭源2024-12-03
401Granite 4.1 3BIBM4.7103.941.2%1.1%11.9%3%-77.383.4%31.4%0%开放2026-04-29
402NVIDIA Nemotron Nano 12B v2 VLNVIDIA4.58*17.6%17%-72.754.5%43.9%0%开放2025-10-28
403MiniCPM-V 4.6 1.3BOpenBMB4.19-108.354.3%0%2.1%6.3%-84.674.9%30.5%0%开放2026-05-11
404Jamba Reasoning 3BAI21 Labs4.13*5.9%7%-73.684.6%33.3%0%开放2025-10-08
405Qwen3 VL 4B历史Alibaba4.09*13.7%13%-75.823.7%37.1%0%开放2025-10-14
406Reka Flash 3Reka AI4.06*26.7%0%-64.75.1%52.9%0%开放2025-03-10
407Olmo 3 7B ThinkAllen Institute for AI3.97*21.2%0%-73.685.7%51.6%0%开放2025-11-20
408Claude 3 Haiku历史Anthropic3.86*18.6%21%-47.653.9%37.4%0%闭源2024-03-04
409Ling-mini-2.0InclusionAI3.75*13.5%6.7%-78.775%56.2%0%开放2025-09-09
410Llama 3 70B历史Meta3.47*18.9%0%-53.584.4%37.9%0%开放2024-04-18
411Llama 3.2 11B (Vision)Meta3.33*11.2%11.7%-63.085.2%22.1%0%开放2024-09-25
412Qwen3.5 0.8BAlibaba3.31-97.241.2%0.4%2.9%6.7%-894.9%23.6%0%开放2026-03-02
413Sarvam M历史Sarvam3.03*17.8%0%-62.253.3%41.6%0%开放2025-05-23
414Exaone 4.0 1.2BLG AI Research2.9*9.3%0%-82.435.8%51.5%0%开放2025-07-15
415Olmo 3 7BAllen Institute for AI2.81*10.3%0%-77.755.8%40%0%开放2025-11-20
416Exaone 4.0 1.2BLG AI Research2.77*7.4%0%-82.585.8%42.4%0%开放2025-07-15
417LFM2.5-1.2B-ThinkingLiquid AI2.75*4.2%0%-83.876.1%33.9%0%开放2026-01-20
418Jamba 1.7 MiniAI21 Labs2.73*9.3%12.7%-74.254.5%32.2%0%开放2025-07-07
419LFM2 2.6BLiquid AI2.71*2.5%0%-52.15.2%29.2%0%开放2025-09-23
420LFM2.5-1.2B-InstructLiquid AI2.71*2.3%0%-73.786.8%32.6%0%开放2026-01-05
421Granite 4.0 H 1BIBM2.66*8.2%6.3%-73.655%26.3%0%开放2025-10-28
422Qwen3 1.7B历史Alibaba2.63*4.3%0%-77.24.8%35.6%0%开放2025-04-28
423Gemma 3 270MGoogle2.41*0%0%-31.324.2%22.4%0%开放2025-08-14
424Apertus 70B InstructSwiss AI Initiative2.4*5.7%0%-55.15.5%27.2%0%开放2025-09-02
425Granite 4.0 MicroIBM2.37*11.9%4%-77.925.1%33.6%0%开放2025-09-22
426Mistral 7B历史Mistral2.14*2.4%0%-64.774.3%17.7%0%开放2023-09-27
427Granite 4.0 1BIBM2.07*8.7%4%-81.825.1%28.1%0%开放2025-10-28
428Molmo2-8BAllen Institute for AI2.03*13.3%0%-69.124.4%42.5%0%开放2025-12-11
429LFM2 8B A1BLiquid AI1.78*6.8%0%-75.134.9%34.4%0%开放2025-10-07
430Granite 3.3 8B历史IBM1.75*10.1%4.3%-76.624.2%33.8%0%开放2025-04-16
431Qwen3 1.7B历史Alibaba1.53*6.9%0%-83.475.2%28.3%0%开放2025-04-28
432Qwen3 0.6B历史Alibaba1.28*2.8%0%-81.785.7%23.9%0%开放2025-04-28
433Llama 3 8B历史Meta1.19*11.9%0%-69.555.1%29.6%0%开放2024-04-18
434Gemma 3n E4B历史Google1.19*0.2%0.7%8.1%0%-80.44.4%29.6%0%开放2025-06-26
435LFM2 1.2B历史Liquid AI1.15*2.5%0%-71.625.7%22.8%0%开放2025-07-10
436Gemma 3 4B历史Google1.12*0.4%0.4%7.3%5.7%-83.325.2%29.1%0%开放2025-03-12
437Llama 3.2 1B历史Meta1.1*1.7%5%-55.225.3%17.7%0%开放2024-09-25
438LFM2.5-VL-1.6BLiquid AI1.01*3%0%-83.875.1%28.9%0%开放2026-01-05
439Granite 4.0 350MIBM1*0.9%0%-72.155.7%20.3%0%开放2025-10-28
440Apertus 8B InstructSwiss AI Initiative1*4.1%0%-74.725%25.6%0%开放2025-09-02
441Gemma 3 1B历史Google1*0.7%0%-77.275.2%23.7%0%开放2025-03-13
442Gemma 3n E2B历史Google1*5.2%0%-79.584%22.9%0%开放2025-06-26
443Tiny Aya GlobalCohere1*3.6%0%-84.725.2%30.5%0%开放2026-02-17
444Qwen3 0.6B历史Alibaba1*4.1%0%-85.535.2%23.1%0%开放2025-04-28
445Granite 4.0 H 350MIBM1*1.7%0%-87.256.4%25.7%0%开放2025-10-28
446Cogito v2.1Deep Cogito41%21.7%-24.9811%76.8%0%开放2025-11-18
447Mi:dm K 2.5 Pro PreviewKorea Telecom29.7%11%-62.028.8%72.2%0%闭源2025-12-11

这个评测关注什么

拆解模型在知识、推理、编程、工具使用和长上下文上的差异。

阅读时注意

评测结果受任务集合、评分方法和模型版本影响。本站保留原始口径,不用它生成新的综合排名。

适用场景

用于回答一个明确的问题:模型在这一类任务上的相对表现如何,而不是“哪个模型绝对最好”。