OnlySOTA

Artificial Analysis LLM Leaderboard

照 Artificial Analysis 发布的原样呈现:它的指标,它的模型。点模型名进本站详情,点箭头回到原榜。

发布方
Artificial Analysis
许可协议
未声明
更新频率
持续更新
本站抓取时间
2026-10-06
条目数
691 · 3 条未识别
按 Artificial Analysis Intelligence Index 排名前 12 的模型
  1. Claude Opus 5.5 57.6
  2. Claude Sonnet 5.5 56.0
  3. Claude Fable 5.1 53.4
  4. GPT 6 Astra 52.7
  5. Gemini 4 Argon 52.6
  6. GPT 6.1 Sol 51.8
  7. Claude Opus 5 50.8
  8. Claude Fable 5 49.6
  9. Muse Spark 1.3 48.1
  10. GPT 6 Sol 47.6
  11. GPT 5.6 Sol 47.0
  12. Grok 4.7 46.4

682 个有分数 · 9 个没测

#模型
1Claude Opus 5.5↗Default · maxAnthropic57.6—61.4%66.9%31.7%——68.3%84.7%46.459.6%41.4%—87.7%$8.0097/s717.98s
2Claude Sonnet 5.5↗Default · maxAnthropic56.0—55.0%61.0%31.4%——67.0%82.7%32.363.6%53.0%——$4.00129/s441.91s
3Claude Opus 5.5↗Default · xhighAnthropic56.0—57.5%65.0%31.7%——66.8%84.7%42.659.6%34.3%—86.6%$8.0080/s131.98s
4Claude Opus 5.5↗Default · highAnthropic53.6—55.6%60.4%30.9%——60.3%82.7%40.656.6%32.4%—85.8%$8.0079/s33.37s
5Claude Fable 5.1↗Default · maxAnthropic53.493.7%59.1%63.1%29.7%47.2%91.4%62.9%85.3%43.552.0%27.4%——$20.0064/s276.79s
6Claude Fable 5.1↗Default · xhighAnthropic53.293.4%58.7%60.9%31.1%45.8%91.0%61.8%83.0%42.455.1%29.5%——$20.0064/s82.80s
7GPT 6 Astra↗maxOpenAI52.796.1%54.7%56.5%31.7%41.4%88.4%52.1%80.7%43.459.1%48.7%—86.9%$20.0063/s343.41s
8Gemini 4 Argon↗highGoogle52.6—57.1%61.8%27.1%——56.3%79.7%42.457.1%84.9%——$4.00——
9GPT 6 Astra↗xhighOpenAI52.496.3%54.6%55.7%31.4%43.1%89.1%50.8%80.0%43.459.6%51.7%—86.2%$20.0059/s199.17s
10Claude Sonnet 5.5↗Default · xhighAnthropic51.9—50.0%57.3%31.1%——61.5%79.7%23.557.1%37.1%——$4.00111/s35.86s
11GPT 6.1 Sol↗maxOpenAI51.8—52.9%54.2%31.7%——53.8%83.0%41.556.1%45.7%—86.0%$4.0059/s308.54s
12Claude Opus 5.5↗Default · mediumAnthropic51.2—54.7%59.3%27.7%——54.3%84.3%40.352.5%31.6%—85.7%$8.0079/s20.64s
13Claude Fable 5.1↗Default · highAnthropic51.290.6%55.9%58.7%30.3%43.1%89.9%56.8%83.7%40.852.0%31.2%——$20.0058/s22.67s
14GPT 6.1 Sol↗xhighOpenAI51.0—52.6%55.7%31.7%——50.5%79.7%40.954.0%49.1%—85.1%$4.0057/s151.76s
15GPT 6 Astra↗highOpenAI50.994.9%53.1%55.4%28.9%40.0%89.9%49.2%80.0%43.754.0%55.2%—86.4%$20.0056/s94.20s
16Claude Opus 5↗maxAnthropic50.893.2%54.9%56.4%29.1%42.1%89.1%61.2%79.3%37.149.0%39.2%—84.7%$10.0054/s57.42s
17GPT 6.1 Sol↗highOpenAI50.2—51.4%55.8%30.0%——49.3%82.3%41.551.5%50.6%—84.9%$4.0052/s62.14s
18Claude Opus 5↗xhighAnthropic49.793.7%54.4%55.7%27.7%43.3%88.0%59.6%80.3%35.446.5%40.5%—84.0%$10.0056/s37.68s
19Claude Fable 5↗Opus 4.8 · maxAnthropic49.692.6%55.5%61.0%28.6%38.1%84.6%55.6%82.3%43.342.4%36.4%63.5%—$20.0068/s112.27s
20GPT 6 Astra↗mediumOpenAI49.693.9%52.7%54.2%29.1%35.5%89.5%48.4%79.7%42.249.5%53.5%—85.1%$20.0051/s5.47s
21Claude Fable 5.1↗Default · mediumAnthropic48.988.6%53.8%56.4%29.1%41.0%88.0%52.5%84.7%37.644.9%30.9%——$20.0055/s6.35s
22Claude Opus 5↗highAnthropic48.193.7%52.8%55.4%28.3%44.7%87.6%54.8%79.0%33.746.0%38.8%—82.4%$10.0056/s21.38s
23Muse Spark 1.3↗maxMeta48.193.5%48.7%58.8%24.9%50.5%84.3%59.2%83.0%25.033.3%67.1%——$2.00187/s47.11s
24GPT 6.1 Sol↗mediumOpenAI47.8—49.9%53.2%27.7%——46.6%83.3%40.048.0%48.4%—83.9%$4.0049/s6.17s
25GPT 6 Sol↗maxOpenAI47.6—47.9%57.6%30.9%——50.5%83.7%27.143.9%39.9%—82.9%$4.00100/s109.84s
26GPT 5.6 Sol↗maxOpenAI47.094.1%49.5%57.1%32.3%44.3%88.0%55.6%84.0%22.039.9%7.8%72.7%83.4%$8.0088/s108.22s
27Claude Fable 5.1↗Default · lowAnthropic46.888.1%48.9%56.7%27.7%39.0%85.0%48.4%82.3%34.140.4%34.4%——$20.0054/s3.57s
28Claude Sonnet 5.5↗Default · highAnthropic46.8—45.8%53.7%24.6%——52.5%78.0%20.943.9%35.4%——$4.00107/s11.94s
29Grok 4.7↗xhighSpaceXAI46.4—43.1%57.4%17.7%——60.8%76.7%32.025.8%70.7%——$3.0088/s60.34s
30Grok 4.7↗highSpaceXAI46.3—42.3%57.8%18.0%——60.5%77.0%30.924.7%67.6%——$3.0096/s72.31s
31MiMo V2.6 Pro↗Xiaomi46.3—49.4%60.9%26.6%——59.3%86.3%8.434.8%59.4%——$0.5447/s5.52s
32GPT 6 Astra↗lowOpenAI45.893.1%49.2%54.1%26.3%32.0%88.0%43.3%80.0%40.541.9%53.1%—84.6%$20.0054/s2.55s
33Qwen 3.8 Max↗0902Alibaba45.492.8%43.1%52.1%17.7%47.8%88.8%58.6%80.3%12.038.9%71.2%—82.8%$3.0038/s2.89s
34Muse Spark 1.3↗xhighMeta45.194.1%47.5%59.7%26.0%47.2%85.4%56.5%83.0%23.116.7%68.5%—82.0%$2.00192/s34.12s
35Claude Opus 5↗mediumAnthropic44.891.9%51.3%51.5%26.9%38.6%86.1%49.6%82.0%31.034.3%39.3%—81.6%$10.0058/s3.50s
36GLM 5.3↗maxZ AI44.891.7%42.3%59.0%19.1%50.3%83.9%57.6%79.7%14.341.9%70.4%——$2.1572/s2.82s
37Grok 4.6↗highSpaceXAI44.394.9%42.9%56.5%17.1%50.7%88.4%56.1%80.3%30.521.2%65.7%——$3.0073/s36.97s
38GPT 6 Sol↗xhighOpenAI44.2—46.3%55.1%28.0%——47.9%81.3%26.730.3%41.1%—82.5%$4.00101/s49.54s
39Grok 4.6↗xhighSpaceXAI44.293.5%44.1%53.0%19.7%43.3%88.0%57.3%81.0%29.317.2%76.0%——$3.0068/s34.43s
40GPT 5.6 Sol↗xhighOpenAI44.093.1%47.3%57.1%28.6%38.1%89.5%53.6%82.3%21.024.7%8.1%71.0%82.7%$8.0091/s42.50s
41Step 5↗previewStepFun43.7—46.5%58.9%20.9%——54.3%88.3%16.433.3%57.0%—76.4%$1.4387/s2.82s
42Kimi K3↗maxKimi43.693.5%46.9%59.5%23.4%46.0%85.0%51.8%88.7%19.712.6%46.8%—80.5%$6.0043/s3.92s
43Grok 4.6↗mediumSpaceXAI42.893.5%42.1%55.9%17.7%44.3%84.3%55.9%81.0%28.013.1%76.0%——$3.0071/s30.57s
44GPT 6 Sol↗highOpenAI42.4—44.1%54.9%25.4%——44.8%83.7%26.826.3%41.9%—82.0%$4.00101/s17.27s
45GPT 5.6 Sol↗highOpenAI42.392.8%46.0%57.8%25.7%36.7%87.3%50.2%81.7%20.420.7%8.8%69.2%81.8%$8.0083/s16.23s
46Claude Opus 5.5↗Default · lowAnthropic42.3—48.3%58.6%17.7%——36.8%80.7%38.931.3%32.4%—84.7%$8.0079/s8.29s
47Grok 4.7↗lowSpaceXAI42.2—39.2%54.9%13.4%——54.4%80.0%29.616.2%60.1%—78.2%$3.0087/s5.07s
48GPT 6.1 Sol↗lowOpenAI42.1—47.4%53.2%24.9%——39.9%84.0%37.630.8%48.4%—83.1%$4.0049/s2.75s
49GPT 5.6 Terra↗maxOpenAI42.192.5%42.9%55.0%30.0%40.2%88.0%47.7%83.0%0.135.4%12.1%71.2%80.7%$4.50118/s126.91s
50GLM 5.3 Flash↗Z AI41.891.2%39.9%51.6%15.4%47.2%84.3%57.3%80.0%7.532.8%72.4%——$0.2452/s3.23s
51Claude Opus 4.8↗maxAnthropic41.892.0%48.7%54.4%20.9%34.2%84.6%47.8%77.7%28.821.7%60.7%62.2%—$10.0064/s52.09s
52Unknown developerUDLing 3.1 Flash未识别41.1—39.4%54.1%18.0%——56.1%83.0%2.233.3%62.1%——$0.45211/s1.80s
53Gemini 3.8 Flash↗highGoogle40.995.3%47.8%56.6%18.3%44.9%87.6%46.8%81.3%29.619.7%44.8%—85.6%$1.50227/s19.72s
54Claude Sonnet 5.5↗Default · mediumAnthropic40.8—39.8%52.9%16.9%——41.2%76.3%20.129.8%48.8%——$4.00105/s7.16s
55Claude Opus 4.7↗maxAnthropic40.7估算91.4%42.3%—12.0%34.6%83.1%42.8%78.7%27.3—57.7%58.6%78.8%$10.0050/s21.47s
56Qwen 3.8 Max↗Alibaba40.292.7%43.0%53.2%20.0%51.3%81.3%55.4%78.3%3.418.7%58.3%—82.3%$3.0037/s2.59s
57Qwen 3.8 Max↗Alibaba39.993.5%42.4%54.1%20.0%49.1%82.0%55.6%80.3%4.311.1%60.8%——$3.0039/s2.79s
58Qwen 3.8 Flash Next↗Alibaba39.892.3%38.0%50.6%11.1%45.4%86.1%56.7%79.7%-9.725.3%54.7%—79.8%$0.2355/s2.41s
59GPT 6 Sol↗mediumOpenAI39.8—41.0%53.8%24.6%——42.5%82.3%27.018.7%43.2%—80.4%$4.00——
60Gemini 3.8 Flash↗mediumGoogle39.893.5%42.1%55.1%12.3%45.8%83.9%46.3%84.0%28.619.7%48.1%—84.2%$1.50——
61Gemini 3.7 Flash↗mediumGoogle39.6估算92.1%39.0%59.8%9.4%35.5%78.3%43.0%83.0%23.7—34.1%—84.7%$1.50285/s6.11s
62Muse Spark 1.2↗xhighMeta39.690.4%45.5%57.4%17.7%34.8%80.1%48.9%79.0%27.27.1%66.7%——$2.00301/s15.17s
63DeepSeek V4.1 Flash↗maxDeepSeek39.5—39.2%51.9%14.3%——55.0%84.0%-5.326.8%3.5%—77.0%$0.52227/s1.08s
64Claude Opus 5↗lowAnthropic39.488.9%43.4%49.2%23.1%30.3%76.4%40.2%81.3%28.626.3%37.8%—79.8%$10.0058/s2.66s
65GPT 5.6 Sol↗mediumOpenAI39.292.6%42.2%57.4%22.9%36.5%86.1%46.1%80.3%19.414.6%9.2%69.6%81.4%$8.0085/s4.80s
66Gemini 3.7 Flash↗highGoogle39.194.5%47.9%57.2%14.3%32.8%85.8%44.6%81.7%26.513.6%35.5%—85.5%$1.50282/s10.81s
67GPT 5.4↗xhighOpenAI39.0估算92.0%43.7%—23.4%39.6%78.3%37.4%82.0%5.8—8.3%73.9%78.4%$5.63107/s150.48s
68Grok 4.5↗highSpaceXAI38.893.1%42.7%55.0%15.4%42.1%81.6%44.5%79.3%25.310.6%45.9%—80.4%$3.0067/s18.78s
69Unknown developerUDMistral Large 4 Previewpreview未识别38.4—35.0%54.2%10.6%——46.2%81.3%-5.326.8%58.1%—76.4%$2.06116/s1.46s
70GPT 5.5↗xhighOpenAI38.493.5%45.8%55.8%27.1%39.0%84.3%42.7%84.3%20.514.6%11.0%75.9%79.9%$11.25104/s43.82s
71Claude Sonnet 5↗maxAnthropic38.291.1%41.3%54.3%16.9%37.3%80.5%48.3%82.0%16.414.1%60.6%—77.3%$4.0086/s230.61s
72GPT 6 Luna↗maxOpenAI38.1—38.5%54.6%19.4%——46.9%83.3%0.712.6%23.3%—79.7%$0.20132/s103.83s
73GPT 5.6 Terra↗xhighOpenAI38.090.8%41.9%52.3%27.1%29.7%80.1%47.5%79.0%-3.010.1%11.0%66.3%79.5%$4.50112/s34.69s
74MiMo V2.6 Flash↗Xiaomi37.9—35.1%51.3%12.0%——55.5%74.3%-12.722.7%45.6%—73.1%$0.1858/s4.96s
75GPT 5.6 Luna↗maxOpenAI37.391.1%39.5%53.6%20.6%31.1%80.9%48.2%83.7%-10.311.6%7.4%—78.6%$0.45128/s101.95s
76GPT 5.5↗highOpenAI37.093.2%45.0%56.1%25.4%36.7%79.4%41.5%84.3%18.89.1%10.9%71.6%81.1%$11.25100/s28.53s
77Gemini 3.7 Flash↗lowGoogle36.9估算90.1%35.1%55.7%5.7%29.5%79.8%41.4%78.7%22.1—32.3%—84.9%$1.50286/s1.32s
78DeepSeek V4 Pro↗0813 · maxDeepSeek36.092.8%41.0%51.0%18.0%39.6%78.7%47.8%80.3%0.814.1%5.2%——$1.9899/s1.66s
79Claude Sonnet 5.5↗Default · lowAnthropic35.9—36.2%49.1%11.4%——33.9%76.0%19.420.7%49.8%——$4.00101/s1.05s
80Grok 4.6↗lowSpaceXAI35.187.9%27.6%49.4%5.7%38.1%75.3%45.5%80.7%25.93.0%69.4%——$3.0066/s7.14s
81DeepSeek V4 Flash Vision↗maxDeepSeek34.891.3%34.5%49.7%10.9%41.0%74.2%52.4%81.3%-17.612.1%8.5%—74.8%$0.66234/s0.97s
82GPT 6 Luna↗xhighOpenAI34.6—34.3%51.7%17.4%——43.2%80.0%-1.88.1%17.6%—79.7%$0.20125/s29.58s
83GPT 5.6 Luna↗xhighOpenAI34.689.5%37.0%50.5%20.6%28.7%77.9%45.1%81.7%-10.83.5%7.5%—78.6%$0.45122/s42.63s
84Claude Sonnet 5↗xhighAnthropic34.4—39.0%54.1%15.4%——43.1%76.7%3.27.1%41.4%——$4.0073/s25.47s
85DeepSeek V4 Flash↗0731 · maxDeepSeek34.390.8%38.6%50.3%16.6%39.4%78.7%46.9%79.7%-14.312.1%8.3%——$0.66227/s0.89s
86GLM 5.3↗lowZ AI34.3—36.6%42.0%14.6%——41.3%72.7%-8.434.8%36.0%——$2.1581/s3.18s
87GPT 5.6 Terra↗highOpenAI34.289.6%38.5%52.4%22.9%28.7%75.7%43.9%77.7%-3.51.5%10.2%64.4%79.1%$4.50100/s2.70s
88GPT 6 Sol↗lowOpenAI34.2—34.9%50.2%16.3%——35.2%79.3%26.59.1%49.3%—80.2%$4.0098/s1.71s
89Gemini 3.6 Flash↗highGoogle34.092.8%40.8%53.4%10.6%29.9%77.5%39.3%80.0%22.17.1%44.4%—83.2%$1.50192/s14.68s
90JT 4.1 Flash↗236B A21B · onChina Mobile33.9—43.6%52.0%14.0%——41.4%78.3%20.63.0%43.0%—————
91GPT 5.5↗mediumOpenAI33.892.6%42.4%54.5%18.6%29.9%80.5%36.7%83.0%18.15.1%10.4%71.0%81.2%$11.2589/s6.42s
92Muse Spark 1.1↗xhighMeta33.789.8%46.2%58.8%15.1%31.8%77.9%35.7%77.7%28.16.1%50.0%——$2.00138/s29.12s
93GLM 5.2↗maxZ AI33.789.5%41.1%51.2%20.9%34.6%77.9%43.7%78.3%4.41.0%73.7%73.3%—$2.1595/s2.96s
94Qwen 3.8↗27B · xhighAlibaba33.790.5%33.9%46.6%5.4%48.0%79.8%46.2%82.0%-10.05.6%69.7%—76.3%$1.1347/s3.83s
95Gemini 3.5 Flash↗mediumGoogle33.6估算92.1%41.3%—10.9%———74.3%20.8—38.2%74.6%83.9%$3.38207/s14.46s
96Motif 3↗Motif Technologies33.6估算83.4%37.0%42.2%6.6%35.3%74.9%31.1%76.0%10.2—71.6%—————
97GPT 5.6 Sol↗lowOpenAI33.589.8%39.4%56.4%14.9%29.1%76.8%40.2%78.0%18.91.0%10.6%66.5%81.0%$8.0075/s2.21s
98Gemini 3.8 Flash↗lowGoogle33.592.0%37.1%55.0%4.0%33.2%83.1%40.5%80.7%21.410.1%35.4%—84.5%$1.50——
99GPT 6 Luna↗highOpenAI32.9—32.9%50.3%15.4%——42.2%79.3%-5.54.5%15.6%—77.5%$0.20134/s20.95s
100Gemini 3.5 Flash↗highGoogle32.692.2%42.7%53.9%13.1%32.2%78.7%35.2%73.3%21.26.6%37.8%76.3%84.3%$3.38213/s16.52s

每一列测的是什么

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
Terminal-Bench 4.0
越高越好
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