跳到正文

模型榜

数据来源:AIHOT

专家难题、研究级数学与物理、逻辑、棋类、做实验反推规律与科学工作流,看模型能不能想明白新问题。AIHOT 评分把各家难度不同的评测折算成同一种分数:50 分是参照模型的平均水平,每 15 分约是一个标准差;两个模型相差 3 分以上,先后基本不会变。评分不是正确率。分类评分只看这个领域,只列在这个领域测得够多的模型。

15 项评测 · 7 家机构 · 更新 2026-11-18 20:05 北京时间

国产由厂商推断,开源依据模型页的权重链接;暂不排名的模型缺少权重标记,开源筛选时不展示。价格单位:¥ / 百万 Token。

名次模型上线日期评测缓存价输入价输出价评分
1Claude Fable 5.2Anthropic2026-11-0911 项评测¥1.33¥53.33¥266.6482.1
2gpt-6.2-astraOpenAI2026-11-179 项评测¥2.67¥53.33¥266.6481.1
3Gemini 4 CarbonGoogle2026-10-286 项评测¥1.33¥26.66¥133.3278.9
4GPT-6 AstraOpenAI2026-09-0315 项评测¥6.67¥66.66¥333.3176.8
5Claude Opus 5.5Anthropic2026-09-2215 项评测¥1.33¥26.66¥133.3275.4
6GPT-6.1 SolOpenAI2026-09-2915 项评测¥0.67¥13.33¥66.6675.3
7gpt-6.2-solOpenAI2026-11-177 项评测¥0.67¥13.33¥66.6674.8
8Gemini 4 HydrogenGoogle2026-11-123 项评测¥0.5¥10¥5074.6
9Claude Fable 5.1Anthropic2026-09-0112 项评测¥1.67¥66.66¥333.3173.3
10GPT-6.1 TerraOpenAI2026-10-266 项评测¥0.33¥6.67¥33.3373.3
11Claude Sonnet 5.5Anthropic2026-09-2815 项评测¥0.67¥13.33¥66.6673.1
12Gemini 4 ArgonGoogle2026-09-3014 项评测¥1.33¥26.66¥133.3272.5
13Claude Opus 5Anthropic2026-07-2412 项评测¥3.33¥33.33¥166.6570.7
14GPT-5.6 SolOpenAI2026-07-0912 项评测¥3.33¥33.33¥199.9870.2
15Claude Fable 5Anthropic2026-06-098 项评测¥6.67¥66.66¥333.3169.5
16GPT-6 SolOpenAI2026-09-2212 项评测¥1.33¥13.33¥66.6669.5
17Kimi K3.1 ProMoonshot AI2026-10-2310 项评测¥1.3¥13¥6869.4
18Muse WonderMeta2026-10-1211 项评测¥1.33¥26.66¥133.3267.7
19GPT-5.5OpenAI2026-04-237 项评测¥3.33¥33.33¥199.9867.7
20Qwen4 MaxAlibaba2026-11-165 项评测¥0.5¥10¥3067.0
21GPT-5.6 TerraOpenAI2026-07-0910 项评测¥1.33¥13.33¥79.9965.7
22Muse EmberMeta2026-11-093 项评测¥0.0267¥0.67¥3.3365.5
23Gemini 3.8 FlashGoogle2026-09-0215 项评测¥1¥10¥5065.3
24Hy4Tencent2026-11-044 项评测¥0.3¥6¥1864.9
25Muse Spark 1.3Meta2026-09-0210 项评测¥1¥8.33¥28.3364.6
26DeepSeek V4.1 Flash 1017DeepSeek2026-10-1711 项评测¥0.02¥1¥464.6
27Hy 4 PreviewTencent2026-08-283 项评测¥0.3¥6¥1864.3
28Gemini 3.7 FlashGoogle2026-08-138 项评测¥1¥10¥5063.8
29Qwen3.8 MaxAlibaba2026-07-1911 项评测¥1.5¥12¥3663.6
30Claude Opus 4.8Anthropic2026-05-288 项评测¥3.33¥33.33¥166.6563.4
31Claude Haiku 5.5Anthropic2026-10-0713 项评测¥0.0667¥0.67¥3.3363.3
32Grok 4.6xAI2026-08-1212 项评测¥3.33¥13.33¥4062.7
33Inkling 2 SmallThinking Machines Lab2026-11-057 项评测¥0.2¥2¥6.6762.7
34Kimi K3Moonshot AI2026-07-1613 项评测¥2¥20¥10062.6
35GPT-5.4OpenAI2026-03-054 项评测¥1.67¥16.67¥99.9962.3
36Grok 4.7xAI2026-09-2115 项评测¥3.33¥13.33¥4061.6
37GLM-5.3Z.ai2026-08-1813 项评测¥2¥8¥2861.5
38Qwen3.8 Max 0902Alibaba2026-09-016 项评测¥1.5¥12¥3661.1
39Claude Opus 4.7Anthropic2026-04-167 项评测¥3.33¥33.33¥166.6560.9
40DeepSeek V4.1 FlashDeepSeek2026-09-1012 项评测¥0.02¥1¥460.8
41Muse Spark 1.2Meta2026-08-055 项评测¥1¥8.33¥28.3360.7
42MiniMax-M3.1-FlashMiniMax2026-10-197 项评测¥0.1¥1¥460.5
43Step 5 PreviewStepFun2026-09-1810 项评测¥0.7¥7¥2060.3
44Claude Sonnet 5Anthropic2026-06-307 项评测¥1.33¥13.33¥66.6660.2
45DeepSeek V4 Pro 0813DeepSeek2026-08-1313 项评测¥0.15¥4.5¥13.560.0
46MiMo-V2.6-ProXiaomi2026-09-218 项评测¥0.025¥3¥660.0
47Muse Spark 1.1Meta2026-07-094 项评测¥1¥8.33¥28.3359.9
48Gemini 3.1 Pro PreviewGoogle2026-02-1910 项评测¥1.33¥13.33¥79.9959.7
49Gemini 3.5 FlashGoogle2026-05-198 项评测¥1¥10¥59.9958.9
50Claude Opus 4.6Anthropic2026-02-057 项评测¥3.33¥33.33¥166.6558.9

一般用量:缓存命中:输入:输出 = 7:2:1。 2026-11-17 · $1 = ¥6.6661

暂不排名 · 5

Muse Wonder FullMeta

本轮未收录官方 API 价格或公开权重

70.6
ember-1Fireworks AI

评测数量还不够

61.8
Qwen3.8 2.4 T A 95 BAlibaba

评测数量还不够

61.4
Ling 3.1 Flash 1019InclusionAI

评测数量还不够

60.4
Ling 3.1 FlashInclusionAI

评测数量还不够

59
全部模型 · 45
名次模型上线日期评测缓存价输入价输出价评分
51GPT-6 LunaOpenAI2026-09-2213 项评测¥0.0667¥0.67¥3.3358.9
52Qwen3.8 Flash NextAlibaba2026-08-265 项评测¥0.1¥0.8¥2.758.8
53GPT-5.6 LunaOpenAI2026-07-0912 项评测¥0.13¥1.33¥858.7
54Grok 4.5xAI2026-07-086 项评测¥2¥13.33¥4058.7
55DeepSeek V4 Flash 0731DeepSeek2026-07-319 项评测¥0.05¥1.5¥4.558.3
56MiMo V 2.6 FlashXiaomi2026-09-217 项评测¥0.02¥1¥258.2
57Muse Spark OpenMeta2026-10-155 项评测———57.8
58Gemini 3.6 FlashGoogle2026-07-219 项评测¥1¥10¥5057.4
59Qwen3.7 MaxAlibaba2026-05-196 项评测¥2.4¥12¥3657.0
60GLM-5.3 FlashZ.ai2026-08-2613 项评测¥0.23¥0.8¥2.856.5
61GLM-5.2Z.ai2026-06-167 项评测¥2¥8¥2854.0
62Mistral Large 4Mistral AI2026-10-066 项评测¥0.47¥4.67¥1453.6
63Beam 1.0Reflection2026-10-168 项评测———53.0
64DeepSeek V4 Pro PreviewDeepSeek2026-04-248 项评测¥0.025¥3¥652.5
65Qwen3.8-27BAlibaba2026-08-148 项评测¥0.6¥3¥1252.2
66Claude Sonnet 4.6Anthropic2026-02-176 项评测¥2¥20¥99.9952.2
67Kimi K2.6Moonshot AI2026-04-206 项评测¥1.1¥6.5¥2752.2
68Kimi K2.7 CodeMoonshot AI2026-06-125 项评测¥1.3¥6.5¥2751.6
69MiniMax-M3MiniMax2026-06-019 项评测¥0.42¥2.1¥8.450.8
70Qwen3.7 PlusAlibaba2026-06-014 项评测¥0.4¥2¥849.4
71DeepSeek V4 Flash PreviewDeepSeek2026-04-243 项评测¥0.02¥1¥249.1
72InklingThinking Machines Lab2026-07-158 项评测¥1.13¥6.67¥2748.3
73MiMo-V2.5-ProXiaomi2026-04-222 项评测¥0.025¥3¥648.1
74Grok 4.3xAI2026-04-175 项评测¥1.33¥8.33¥16.6747.5
75GPT-5.4 Mini 2026-03-17OpenAI2026-03-177 项评测¥0.5¥5¥3047.2
76Inkling-SmallThinking Machines Lab2026-07-307 项评测¥0.4¥2¥847.1
77GPT-5.4 Nano 2026-03-17OpenAI2026-03-177 项评测¥0.13¥1.33¥8.3346.8
78GLM-5.1Z.ai2026-04-074 项评测¥1.3¥6¥2446.1
79Nemotron 3 UltraNVIDIA2026-06-045 项评测———46.0
80Qwen3.6 PlusAlibaba2026-04-016 项评测¥0.2¥2¥1245.6
81Nemotron 3.5 Super VLNVIDIA2026-10-155 项评测———45.5
82MiMo-V2.5Xiaomi2026-04-222 项评测¥0.02¥1¥243.5
83Qwen3.6-27BAlibaba2026-04-225 项评测—¥3¥1842.5
84MiniMax M 2.7MiniMax2026-03-182 项评测¥0.42¥2.1¥8.441.2
85Gemini 3.5 Flash LiteGoogle2026-07-2110 项评测¥0.2¥2¥16.6740.6
86Muse GlimmerMeta2026-08-104 项评测———39.3
87GPT-5 MiniOpenAI2025-08-076 项评测¥0.17¥1.67¥13.3338.3
88Gemma 4 31 BGoogle2026-04-024 项评测免费免费免费37.6
89Gemini 2.5 ProGoogle2025-06-174 项评测¥0.83¥8.33¥66.6634.5
90Gemini 3.1 Flash Lite PreviewGoogle2026-03-034 项评测¥0.17¥1.67¥1030.9
91Mistral Medium 3.5Mistral AI2026-04-294 项评测¥1¥10¥5029.9
92Claude Haiku 4.5 20251001Anthropic2025-10-152 项评测¥0.67¥6.67¥33.3328.5
93gpt-oss-120bOpenAI2025-08-055 项评测———25.9
94Granite 4.2 8 BIBM2026-08-254 项评测———20.8
95Granite 4.2 3 BIBM2026-08-254 项评测———10.5